JP2947566B2 - In-frame area detection device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION [Object of the Invention] (Field of Industrial Application) The present invention relates to a process for extracting an area specified by an operator on an image to facilitate editing of the image.

(Prior Art) In recent years, with the progress of digital technology, devices that handle images by digital signals, such as G3 facsimile, document files, and digital copying machines, have been increasing. In such devices, by handling images as digital signals, processing such as image quality improvement and correction, image accumulation and transmission, and editing such as density conversion and area movement according to the operator's will are performed. There is an advantage that processing can be easily performed.

The image editing process is to create a new image according to the operator's intention by moving a part or all of the existing image, converting the size, density, color, and the like. An apparatus that performs such processing is desired to be easy to use in order to perform various processing according to the will of the operator.

The most important factor in performing image editing is area designation. Specify the area with this, color replacement, masking,
By combining simple processing such as trimming, image binarization and movement, various editing processes are possible,
It is a very general processing element. As a method of specifying an area for an image read into the memory, an operator can specify a position using a cursor or a mouse, and interactively specify an area while confirming the position through a console or the like. .

On the other hand, since the image memory is expensive, it is desired that a device for ordinary office use or the like performs processing in real time without writing an image in the memory. In such a device, it is preferable to specify an area when reading an image. In this case, as a method of specifying the area, there is a method of inputting the coordinates of the end point of the area using a tablet or numerical values using a numeric keypad before reading the image. But,
This method has disadvantages such as that the shape of the designated area is limited to a simple figure such as a rectangle, and that the correspondence between the coordinate values given by the numeric keypad and the tablet and the position on the document is difficult to understand, and the usability is poor.

Therefore, in addition to the above method, a method of specifying an area by a frame line is known. In this method, a frame representing the boundary of a region is created, and a region inside the frame is specified. In this method, it is necessary to distinguish between the frame line and the document information. To do this, add a line with different color and density to the original information on the original as a frame line on the original, or prepare a transparent sheet separately from the original, overlay this transparent sheet on the original, and Is formed, and the transparent sheet is read before reading the original image. The method of specifying by the frame line is that the former is limited to colors and densities where the frame line is not used in the original image, the latter increases the time and effort of sheet input, and requires a memory for storing area information, However, there are major advantages such as the ability to specify a free-form area and the ability to specify the specified area on the same plane as the original image, so that the correspondence can be easily understood.

As a method of detecting a region in a frame from a frame line, a method of once writing frame line information in a memory and determining an area by referring to the contents of the memory by a CPU or the like is known. According to this method, complicated frames can be handled,
It is necessary to have a memory that stores the frame image as it is,
Further, there is a drawback that real-time processing is impossible and processing time is required depending on the accuracy of the processing.

Therefore, a method of sequentially detecting the in-frame area in real time from the read raster scan signal is also known. One method is to check connectivity from the outermost area in the range of information up to the raster. In this method, first, a frame portion and a ground portion are identified from an image, and the outermost ground portion is set as an outer frame portion. Then, of the ground portion of the current raster, a portion adjacent to the portion outside the frame of the previous raster and a portion connected to this portion in the main scanning direction are defined as a portion outside the frame of the current raster. Here, the outside portion of the frame is an outside region, and the ground portion other than the outside portion is an inside region of the frame.

According to this method, the area within the frame can be detected from the raster scan signal by simple processing. However, the frame lines that can be detected are limited to topographically simple figures, for example, as shown in FIG. 24, cannot have a concave portion on the lower side, or cannot correspond to a complicated shape such as a double frame. .

(Problem to be Solved by the Invention) As described above, the conventional in-frame area detection method for detecting the in-frame area from the raster signal in real time can be applied only to a very simple topographic frame shape. For example, a double frame or a frame having a concave portion on the lower side has a disadvantage that a detection error occurs partially.

The present invention has been made in view of such a problem, and an object of the present invention is to provide an in-frame area detection device that can always accurately detect an in-frame area even with a complicated frame line.

[Structure of the Invention] (Means for Solving the Problems) An in-frame area detection device according to the present invention includes an identification unit for identifying an image signal obtained by raster scanning into a frame portion and a ground portion other than the frame portion. Storage means for storing respective levels of a frame run which is a connection area of the frame part of each raster of the image signal identified by the identification means and a ground run which is a connection area of the ground part of the image signal; The determined levels are used as the respective levels of the frame run and the ground run of the previous raster, and the level determining means determines the respective levels of the frame run and the ground run of the current raster with reference to the levels. With reference to the levels of the frame run and ground run of the current raster
Determining means for determining whether the ground portion of the current raster is inside or outside the frame.

That is, in the present invention, the level attribute is given to the ground portion or both the frame portion and the ground portion detected from the image. And
The level is determined based on the connection relationship with the previous raster, and the depth of the nest of the ground or frame is detected according to this level.

The signal may be thinned by prioritizing the frame portion before determining the level, or the image signal may be handled in units of runs when determining the level.

(Operation) According to the present invention, by determining the level of the frame and the ground, it is possible to detect the nest level of the frame, that is, the level of the frame with respect to the outermost frame, and thereby the double frame or the downward concave portion. For such a frame, it is possible to detect a reasonable area, and to remove the influence of a beard or the like generated when a handwritten frame is formed.

In addition, by thinning out the signal by giving priority to the frame portion, the signal amount can be reduced without losing the frame line due to the thinning, and the processing can be simplified. In addition, when the level is determined, the processing of the level determination can be simplified by treating the image signal in units of runs.

Hereinafter, a color image editing apparatus according to an embodiment of the present invention will be described with reference to the accompanying drawings.

FIG. 1 is a diagram showing a configuration of a color image editing apparatus according to a first embodiment of the present invention. This apparatus reads a monochrome document image in which an area to be colored, masked, trimmed, or the like is specified by a frame of a specific color, and outputs an image in which the specified area has been subjected to a specified process. The apparatus includes an image input unit 301 as shown in FIG.
A color determination unit 302, a thinning processing unit 309, an in-frame area detection unit 303,
It comprises an image editing unit 304 and an image recording unit 305. Hereinafter, the function of each part will be described in detail.

<Image Input Unit 101> The image input unit 101 reads a document image as an RGB digital image signal. The image input unit 101 includes, for example, a dot-sequential color line sensor in which RGB color filters are sequentially arranged on each light receiving surface of the line sensor. While reading image information line by line, the image input unit 101 is perpendicular to the longitudinal direction of the sensor. The reading scan is performed in the direction. As a result, the original image can be read as a raster scan signal. Hereinafter, the line direction is referred to as a main scanning direction, and the direction perpendicular thereto is referred to as a sub-scanning direction. The image input unit 101 converts an analog signal photoelectrically converted by the line sensor into a digital signal by an A / D converter, and then performs shading correction. This processing is for correcting the variation of the gain and offset of each element of the line sensor and correcting the white balance of the RGB signal, and the image signals become 0 and 1 for the black and white pixels, respectively. It is standardized as follows. For the specific configuration and mechanism of this part, see, for example,
It is described in detail in -71764.

The image input unit 101 further performs a point-sequential correction process on the signal subjected to the shading correction. In a point-sequential color sensor, color information of three slightly shifted points is color-separated, so that color noise occurs at an edge portion on an image. Therefore, the image input unit 101 removes this noise by performing an interpolation process according to a point shift by a point sequential correction process. The details of this processing are described in, for example, JP-A-61-154357.

<Color Discrimination Unit 102> Next, the color discrimination unit 102 discriminates the achromatic original image component on the document from the chromatic frame line. In the present embodiment, identification is performed using a look-up table (LUT). As is well known, when colors are expressed in Lab space, achromatic colors such as white and black become points on the achromatic axis where a = b = 0, and pure colors such as red and blue used as frame lines are achromatic axes. Away from the point.
Therefore, the color can be identified by dividing the Lab space as shown in FIG. In this embodiment, six colors of red, green, blue, black and white are identified as shown in the figure. Here, signals of four colors of red, green, yellow, and blue represent frames, and black represents image information to be originally edited. Then, a frame signal indicating whether the pixel is achromatic, red, green, blue, or yellow and a black signal indicating whether the pixel is black are output from the color determination unit 102. Here, the frame signal represents colors by codes, and codes 0, 1, 2, 3, and 4 are assigned to achromatic colors, red, green, blue, and yellow, for example.

Here, the LUT can be configured by converting the division plane in the Lab space into the RGB space. Alternatively, the LUT may be referred to after converting the RGB signal into a signal system such as a Lab signal.
As a result, the quantization efficiency of the LUT can be increased.

A color-identified signal contains many identification errors due to noise or the like. If the color frame on the document has color unevenness or noise, a frame may be formed in a portion other than the frame line, or a hole may be formed inside the frame line. When such noise occurs, a detection error may occur in frame detection. Since this noise has a small area, the color discrimination unit 102 suppresses these noises by performing processing such as removal of isolated points after color discrimination.

<Thinning Processing Unit 103> The thinning processing unit 103 thins out the pixels in the main scanning direction with respect to the frame color signal 113 identified by the color determining unit 102. This aims to reduce the amount of information of the image signal and reduce the amount of calculation in the in-frame area detection unit 104 described later. The thinning rate is preferably, for example, 1/4. That is, three of the four pixels are thinned out and reduced to one pixel. Further, instead of performing simple thinning-out of selecting and outputting one pixel at a specific position among the four pixels, thinning-out is performed by giving priority to chromatic colors. That is, in FIG. 3, W, R, and GB are achromatic, red,
Assuming that green and blue codes are represented, as shown by A in FIG. 3, if there is a chromatic (red, green, blue or yellow) signal in at least one of the four pixels, the output of each of the thinning-out signals is obtained. I do.
Then, as shown by B in FIG. 3, an achromatic color is output only when all four pixels are achromatic. Also, as shown by C in FIG. 3, when there are two or more frame colors among the four pixels,
The foremost color is thinned out. However, since it is rare to draw different color frame lines densely in a narrow portion, a state as shown in C in FIG. 3 does not usually occur. As described later, in the present embodiment, the one-raster signal is once stored in the memory by the in-frame area detecting unit 104. Therefore, the capacity of the memory can be reduced by thinning out the signals. Further, since the signal transfer speed is reduced, time-division processing can be performed, and the scale of the processing circuit can be reduced. On the other hand, since the line of the frame color is thinned out in preference to the white background, the thin frame line is not faded or lost due to the thinning, and the accuracy of the in-frame area detection accuracy is very little deteriorated. Note that, as described above, the thinning processing unit 103 is for the purpose of reducing the processing scale, and in some cases,
This processing can be omitted.

<In-frame area detection unit 104> The in-frame area detection unit 104 identifies the in-frame area from the frame color signal output from the thinning processing unit 103. First, an algorithm for detecting a region in a frame according to the present embodiment will be described. In the present embodiment, four colors are independently used as types of frame colors, but the case of one color will be described here for convenience.

FIG. 4 shows the algorithm of this processing. In this algorithm, each frame pixel is given a frame level attribute and a ground pixel is a ground level attribute, and the frame level or ground level of the current raster is determined from the frame signal of the previous raster, the frame level, the ground level, and the frame signal of the current raster. Depending on the ground level, either inside or outside the frame is determined. Then, by sequentially repeating this process while proceeding for each raster, the in-frame region of the entire image is detected. Here, since there is no raster before the first raster which is the first raster of the image, the zeroth raster is considered virtually. It is assumed that the 0th raster is all ground and its ground level is an initial value 0. The 0th raster is considered to be the end of the image, and the ground level 0 here is a reference value.

Next, the frame level or ground level of each raster is determined. The procedure for this determination will be described in detail with reference to FIG. The frame signal of the raster is P (x), the frame signal, frame level, and ground level of each pixel of the previous raster are P0 (x),
LF (x) and LG (x). Here, the frame signal is a binary signal that takes 1 in the frame portion and 0 in other portions as described above, and x represents a position in the main scanning direction.

First, the frame signal of the current raster is converted into a run expression. That is, a connected area having the same frame signal value in each raster is called a run, and one raster is divided for each run. Hereinafter, the n-th run is denoted as Rn. Each of these runs represents a frame and the ground, respectively. Next, the ground level of the ground run Rn is determined by the following rules.

If there is no run on the ground adjacent to the run Rn in the previous raster, the ground level of the run Rn is obtained by adding 1 to the frame level of the run adjacent to the previous raster [FIG. 6 (a)]. ].

When there is only one run on the ground adjacent to the run Rn in the previous raster, the ground level of the run is set to the ground level of the run Rn (FIG. 6B).

When there are a plurality of runs of the ground adjacent to the run Rn in the preceding raster, the minimum value among the ground levels of the plurality of grounds is set as the ground level of the run Rn [FIG. 6 (c)].

Further, the frame level of the frame run is determined by the following rules and the like.

If there is no run in the frame adjacent to the run Rn in the previous raster, the ground level of the run adjacent to the previous raster is set to run Rn.
[FIG. 6 (d)].

When there is only one run of the frame adjacent to the run Rn in the previous raster, the frame level of that run is set to the frame level of the run Rn. [FIG. 6 (e)].

When there are a plurality of runs of the frame adjacent to the run Rn in the previous raster, the maximum value among the frame levels of the plurality of frames is set as the frame level of the run Rn [FIG. 6 (f)].

Note that these rules and the conditions a., B.,
c. and d. correspond to a., b., c., and d., respectively. Although FIG. 5 does not distinguish between and, they are classified here for convenience of the following description. However, they have essentially the same meaning.

Here, that the run of the previous raster is adjacent to the run of the current raster means that any of the pixels constituting the current raster is adjacent to any of the pixels of the run of the previous raster. Further, the pixel adjacent to the pixel P (x) of the current raster means only the same pixel P0 (x) at the main scanning position in this embodiment as shown in FIG. 7 (a). However, the adjacent relationship is not limited to this definition. For example, for the pixel P (x) of the current raster, the pixel P0 (x) at the same main scanning position of the previous raster and the pixel P0 (x -1) may be defined as adjacent pixels [FIG. 7 (b)], or three pixels including the next pixel P0 (x + 1) may be defined as adjacent pixels [FIG. 7 (b)]. Figure (c)]. In particular, in the second definition, when the frame and the ground intersect with each other between the previous raster and the current raster, one of the frame and the ground is always connected, and the other is not connected. However, since such a fine frame line is rarely formed, it is not practically a problem even if the adjacent relation is used as shown in FIG. 7A as in this embodiment. Few.

From the value of the ground level determined by the above-described rule, it is possible to determine whether the ground portion is an in-frame area or an out-of-frame area. Since the pixels of the same run have the same level, the portions connected in the main scanning direction have the same level. Also, the rules and the parts connected in the sub-scanning direction according to the rules are at the same level. Therefore,
The ground portion and the frame portion connected in the main scanning direction or the sub-scanning direction in the downward direction inherit the same ground level or frame level [FIG. 8 (a)]. When a new frame is formed from the ground portion (FIG. 8B), the ground level of the adjacent ground portion becomes the frame level of the newly generated frame according to the rules. When a new ground portion is generated from the frame portion [FIG. 8 (c)], a value obtained by adding 1 to the frame level of the frame becomes the ground level of the newly generated ground portion according to the rules. Therefore, when a newly convex frame line is generated, the ground level of the portion surrounded by the frame line [FIG. 8 (d)] is a value obtained by adding 1 to the ground level of the ground portion outside the frame line. That is, the ground level inside the multiply nested frame increases by one as the nest becomes deeper. Therefore, if the inside of a double or more frame line is alternately regarded as inside or outside the frame, it can be determined whether the ground level is an even number or an odd number, and whether the area is within the frame or outside the frame. Since the ground level of the outermost part of the image is 0, it is determined that even-numbered ground levels are outside the frame and odd-numbered areas are inside the frame. In this embodiment, in the rule, the maximum value of the frame level of the frame adjacent to the previous raster is set as the frame level of the frame. However, this rule is not essential, and a system in which the rule is rewritten to the following rule 'as a modified example is also possible.

'If there are a plurality of runs of the frame adjacent to the run Rn in the previous raster, the minimum value among the frame levels of the plurality of frames is set to the run Rn
Frame level.

In the algorithm of this modified example, when the phase relationship between the frame lines is special, an area determination result different from that of the present embodiment is obtained.
However, the advantages and disadvantages of the determination result according to the present embodiment have advantages and disadvantages depending on the phase of the frame line, and there are very few such special cases in a normal frame line. There is no.

Next, a specific example of the in-frame detection by the above-described in-frame area detection algorithm will be described.

(1) Case of Circular Frame First, a case of a simple circular frame as shown in FIG. 9A will be described. In the following drawings, the sub-scanning direction is downward from the top of the drawing. The ground level remains at the initial value 0 because the ground is all ground up to raster 901 in the figure. Raster 901
It's the first time that the border is on. FIG. 9B is a schematic diagram in which this portion is enlarged to the raster level. In the figure, the hatched portion represents the frame line portion, and the other portions represent the ground portion. For the first time in the raster 901, a frame run 902 occurs. Run 90
Since the front raster of No. 2 has no frame portion, the frame level of the run 902 becomes 0 according to the rules of the algorithm. The frame levels of runs 903 to 904 of the subsequent raster frames are set to 0 by inheriting the frame levels of the runs 902 according to the rules. Next, the raster 905 on the ground surrounded by the frame for the first time
Then, a ground run 908 occurs between the run 906 and the run 907 of the frame. run
By rule, the 908 ground level is 1 per run 904 frame level.
, That is, 1, and it is detected that it is inside the frame. In the following rasters, this level is inherited by rules. This allows the area inside the border to be
910 are all determined to have a ground level of 1, that is, an area inside the frame, and all the outer areas 911 of the frame line are determined to have a ground level of 0, that is, an area outside the frame, and the inside and outside of the frame are correctly determined.

(2) In the case of a frame line having a downward concave portion Next, a case of a frame line having a downward concave portion as shown in FIG. 10 (a) will be described. As in the previous example, the ground level of the area 1012 in the frame is set to 1 for the first time until the raster 1001 corresponding to the frame line, and the raster 1002 corresponding to the frame of the downward concave portion for the first time. For the first time, a run 1003 of the frame line of the downward concave portion occurs on the raster 1002 [FIG. 10 (b)]. Since this run is adjacent only to the ground at the ground level 1, it is set to the frame level 1 according to the rules. Then, the ground level of the run 1004 where the ground outside the downward concave portion first appears is 2 by the rule. Therefore, the ground level of the area 1010 of the downward concave portion is 2 up to the raster 1005. Raster
At 1005, the area 1010 of the downward concave portion and the outside ground area 1011 are combined, but the ground level after this raster becomes 0 according to the rule. As a result, the ground level is 0 for all the areas inside the frame.
Or it becomes 2, and inside and outside of the frame are correctly determined.

(3) In the case of an n-folded frame To specify a donut-shaped area shown by the hatched portion in FIG. 11 or an area having a further isolated island in the hole of the donut by using a frame line, FIG. It is reasonable from a geometrical sense and a common sense to specify by a double or triple border. In the present algorithm, as shown below, the inside and outside of the region can be correctly determined in the above-described meaning even for such a double or more frame line. That is, in the case of a triple frame as shown in FIG. 13, the frame level of the outermost frame 1301 is 0, and the ground level of the inner ground portion 1302 is 1
Becomes Similarly, the frame level of the second frame 1303 from the outside is 1,
The ground level of the ground portion 1304 inside is 2. Thus, as the number of nests increases, the level of the frame and the level of the ground increase by one. As a result, the innermost, that is, the inner 1306 of the third frame 1305 from the outside, the second frame from the outside
The ground level of the portion 1304 between 1303 and the second outermost frame 1305, and the ground level of the portion 1302 between the outermost frame 1301 and the second outermost frame 1303 are 3, 2, and 1, respectively, and are the areas within the respective frames. , Outside the frame and inside the frame. This is not limited to the case of a triple frame line, and the same determination is made even if the number of frame line nests increases to four or more.

(4) When there is a beard By the way, in each of the above examples, the frame line is a single closed curve having the same phase as the circle, but when the frame line is drawn by hand, the start point and the end point of the frame line are used. May not match, and the frame may become an open curve as shown in FIG. 14 or a closed curve with a whisker branching inward or outward as shown in FIGS. 15 (a) and 15 (b). It is difficult to draw a closed curve so as not to be in any of these states, which significantly impairs usability.
According to the present algorithm, the area inside the frame can be detected for the lower open frame line as shown in FIG. 14 (a) due to the directionality of the sub-scan, but it can be detected at the upper side as shown in FIG. 14 (b). An area inside the frame cannot be detected with an open frame line or the like. That is, it is not always possible to detect the in-frame area with respect to the open curve.

However, if the curve is a closed curve as shown in FIGS. 15 (a) and 15 (b), even if there is a beard, the inside and outside of the correct frame line are determined. That is, when there is a beard inside the closed curve as shown in FIG. 15 (a), the frame level is 0 in any part, and the ground level inside the closed curve is all 1, that is, the area within the frame. When the closed curve has a beard on the outside as shown in FIG. 15 (b), the frame levels of the frame lines are all zero. Then, the ground level of the inner portion 1501 of the closed curve and the portion 1502 sandwiched between the beard and the closed curve is 1. However, the portion sandwiched between the closed curves is connected to the ground portion having the outer ground level of 0 on the raster 1503 which is the end point of the beard. According to the rule of the algorithm, the smaller the ground level becomes 0 after the raster 1503. For this reason, the inside 1501 of the closed curve and the area 1502 sandwiched between the beard and the closed curve are determined as the in-frame areas. It is common sense to think of it as an area outside the area of the area 1502,
Since the beard is small, the area 1502 is a sufficiently small area, and it is considered that there is no practical problem with such a determination result. It is sufficient to draw the frame line while paying attention so that the frame line becomes a closed curve, and there are few restrictions on the drawing of the frame line, which is convenient.

Next, the in-frame area detection unit 10 for realizing the above algorithm
The specific configuration of 4 will be described. FIG. 16 shows an example of the configuration of the in-frame area detection unit 104. The frame color signal 114 output from the thinning processing unit 103 is decoded by the decoder 1641, and four color frame signals 1651 to 1654 indicating the presence or absence of red, green, blue, and yellow, respectively.
Create These frame signals 1651 to 1654 are supplied to the level determination units 1601 to 1604 and the memory unit 16 provided for each color.
05 to 1608 are input. These four level determination units 16
01 to 1604 and the memory units 1605 to 1608 all have the same configuration. Therefore, here the red frame signal 16
The level determination unit 1601 and the memory unit 1605 corresponding to 51 will be described.

(I) Level Determination Unit 1601 The level determination unit 1601 determines the level of the current raster from the frame signal 1651 of the current raster, the frame signal 1655 of the previous raster, and the level signal 1656 of the previous raster. The determination is made in units of the run of the current raster, that is, the connected frame portion and the connected ground portion. The determination unit 1601 includes a determination logic circuit 1609, a 6-bit internal state register 1610, and an output register 1611. The decision logic circuit 1609 receives the frame signal 1651 of the current raster, the frame signal 1655 of the previous raster, the level signal 1656 of the previous raster, and the output signal 1660 of the register 1611, and receives an internal state consisting of a 2-bit connection state and a 4-bit level state. A signal 1657 and a 1-bit run end signal 1658 are generated. The internal state register 1611 stores the internal state signal 1657 in synchronization with each pixel. As a result, a state transition circuit having four states S0, S1, S2, S3 based on the connection state 2 bits and a 4-bit level state LS is configured.

FIG. 17 shows the input / output relationship of the decision logic circuit 1609. This represents a state transition of the level determination unit. Here, the internal state LS of the pixel for which the run end signal is 1, that is, the pixel whose current raster changes from the frame to the ground or from the ground to the frame, is stored in the output register 16
11, which represents information such as the run level. This level signal becomes the ground level or frame level of this run.

Here, S0, S1, S2, and S3 of the internal state ST are respectively the current raster is the ground and the previous raster has the ground pixel, the current raster is the frame and the previous raster is all the ground, the current raster is the ground and the previous raster is all the frame. , The current raster is a frame, and the previous raster has pixels of the frame. These internal states are determined at the end of the run, and by making the state transitions as shown in FIG. 17, the states S0, S1, S2, and S3 at the end of each run are rules, rules, rules, and rules. And an algorithm for level determination corresponding to the rule is executed in accordance with the condition of the run of the rule. As described above, the algorithm for determining the level can be realized with a simple configuration.

(II) Memory unit 1605 The level signal determined by the level determination unit 1601 is defined in units of runs, and the level value is determined at the last pixel of the run. For this reason, the level signal for one raster determined by the level determining unit 1601 is stored in the memory unit 1605 at the next stage, and is read out as a pixel-based signal whose level is determined from the first pixel of the run in the next raster.

The memory unit 1605 includes two memories 1612 and 1613, an address counter 1614 for reading, an address counter 1615 for writing, a reading register 1616, a comparator 1617, a pixel position counter 1618, a switch 1619, and a logic circuit 1620. I have.

The write operation is performed as follows. First, the first write address counter 1615 of each raster is cleared to the initial value 0. Each time the last pixel of a run is reached, the frame signal of the run, the level signal, and the position coordinates of the last pixel of the run are written to one of the memories 1612 or 1613 as one word, and the write address counter 1615 is incremented by one. I do. Here, the pixel position counter 1618 clears at the beginning of the raster and counts up for each pixel in the main scanning direction to output a signal indicating the pixel position in the main scanning direction. What is necessary is just to write the content of this counter 1618 as the last pixel position of the run. Each time one run is completed, these operations are repeated until the end of the raster, and the last writing is performed with the last pixel of the raster. Through this series of operations, frame signals of all runs on one raster, level signals, and information on the last pixel position are written in the memory 1612 or 1613 in ascending order of the run position.

Then, while reading this information with the next raster, it outputs as a signal in pixel units. First, at the beginning of each raster, the read address counter 1614 is cleared to the initial value 0, and the contents of the memory 1612 or 1613 are written to the read register 1616. In the read register 1616, the frame signal and level of the first run and the last pixel position of the run are written. Then, the comparator 1617 compares the content of the pixel position counter 1618 indicating the pixel position in the main scanning direction with the last pixel position of the run of the readout register 1616. When the comparison result becomes equal, the read address counter 1614
, The contents of the memory are read out and written into the register, and this operation is repeated until the end of the raster. Through this series of processing, the frame signal and the level signal are output from the read register 1616 in synchronization with each pixel.

Here, it takes time for two rasters to write and read a signal of one raster, so two memories are used, one is used for writing, and the other is used for reading. By alternately switching the write operation, signals of all rasters can be processed without interruption.

In this embodiment, two memories are used. However, for example, one memory can be used in a time-division manner.

As described above, since the portion where the ground level is an odd number is the region within the frame, the logic circuit 1620 outputs the least significant bit of the level signal 1661 output from the read register 1616 and the frame signal.
The negation of 1662, that is, the logical product with the ground signal, is taken, and this output is assumed to be the in-frame signal 1671. However, in the present embodiment, the frame portion is regarded as being outside the region, but it is also possible to consider the frame portion as being within the region. In this case, the logical sum of the least significant bit of the level signal and the frame signal may be used as the in-frame area signal. The frame signal and the level signal output from the read register 1616 are used by the level determination unit 1601 as a reference signal of the previous raster.

As described above, by determining the level in units of the run, the determination result determined at the end of the run can be reflected on the pixels of the entire run. This is equivalent to pre-reading information by an arbitrary number of pixels in the direction opposite to the main scanning direction, whereby the level can be inherited in both directions in a region connected to the main scanning direction. . In a pixel-by-pixel process for a normal raster scan signal, if the number of prefetched pixels of a signal is increased, the circuit scale becomes enormous, so the number of prefetched pixels is limited to several pixels. Also, if one raster signal is stored in a memory and the level is determined collectively by using a CPU or the like, high speed is required for the access to the memory and the calculation for determining the level.

However, by performing the level determination by converting to the run expression as in the present embodiment, it is possible to determine the level of the raster scan signal without direction dependency in the main scanning direction with a simple configuration.

In addition, by storing the level information determined in this way for the entire screen in the run expression, the frame area information can be stored even after the entire screen is scanned.
By storing in the run expression, the frame area can be saved with a small memory capacity.

By using this information, it is possible to realize a so-called inset synthesizing process or the like in which the second image scanned afterwards is inserted into the frame portion of one image.

(III) Encoder 1631 By performing the above processing independently for each of the four frame colors, the four-color in-frame signals 1671 to 1674 are obtained. The encoder 1631 determines the color of the area from these signals, and outputs this as the area color signal 115. Since the in-frame signals 1671 to 1674 detect the in-frame regions independently for a plurality of color frames, there is a possibility that an area inside the plurality of color frames may occur. Therefore, one color is determined from the plurality of colors by the encoder 1631 which is an area color determination unit. In this embodiment, priorities are assigned to the color frames, and the region color is determined by the following algorithm. That is, first, an area where all the signals are 0 in the four-color in-frame signals is an area that does not fall within any of the color frames, and this area color is defined as black for convenience. Also,
When only one of the four in-frame signals is 1, the frame color of 1 is set as the region color. When two or more frame signals become 1 at the same time, a frame color having a higher priority among frame colors that become 1 is set as an area color. In this embodiment, the input / output relationship is set in the encoder 1631 as shown in FIG. 18 with the priority order of the frame colors in the order of red, green, blue, and yellow, and is converted into the area color signal 115.

In the present embodiment, when a plurality of in-frame areas overlap, the color is determined by the color priority relationship. However, this is not limited by the present invention. For example, the priority order is determined by the inclusion relationship between different color frames. There is no hindrance to taking such measures as determining

<Image Editing Processing Unit 105 and Image Recording Unit 106> Next, the image editing processing unit 105 performs editing processing specified by the operator on the in-frame area according to the in-frame signal detected by the in-frame area detection unit 104. . In the present embodiment, a case will be described in which color replacement processing and masking processing are performed as editing processing. The color replacement process replaces the black portion of the in-frame region with a designated color, and the masking process deletes the black portion of the in-frame region (replaces with white). This processing can be realized by, for example, a rewritable color replacement table that receives a black signal 352 and an area color signal 355 as input and outputs a recording signal.

The operator can specify the relationship between the frame color and the processing from the control panel, and rewrites the content of the color replacement table according to the specified content. For example, if the area inside the red, green, and blue frame colors is replaced with the same color as the frame color, and the area inside the yellow frame color is masked, the table Are set, for example, as shown in FIG. That is, when the black signal is 0, white is output as a recording signal, and when the black signal is 1 and the area color signal is red, green, and blue, a recording signal representing red, green, and blue is output, When the black signal is 1 and the area color signal is yellow, a recording signal representing white is output. When the black signal is 1 and the area color signal is black, a recording signal representing black is output. Finally, according to this recording signal, the image recording unit 30
In step 5, an image is recorded on the recording paper. Thereby, for example, an image [FIG. 20 (b)] in which a color frame is added to a monochrome image as shown in FIG. 20 (a) is input, and an edited image as shown in FIG. 20 (c) is input. Can be recorded. That is, a color replacement process, a masking process, and the like designated by the operator in correspondence with each frame color can be realized.

In the present embodiment, only the masking process is performed as the color replacement process in the designated frame as the editing process. However, the present invention is not limited to such a process. Can be applied to an inset synthesizing process in which an image of another portion is moved.

With the above configuration, it is possible to realize a color image editing apparatus that performs editing processing such as color replacement in accordance with a color frame added to a monochrome original. In particular, by defining a frame and other levels using the present invention, an area surrounded by a frame line of double or more and a frame line of a downward concave portion can be correctly detected, thereby realizing an image editing device which is easy for the operator to use. it can.

In this embodiment, pipeline processing is performed on a raster scan image signal without using an image memory. However, the present invention is not limited to the application to such pipeline processing. For example, while reading the images stored in the image memory one raster at a time, the in-frame area detection processing shown in the present embodiment is performed. The present invention is also applicable to an apparatus in which the detected in-frame area information is written into the image memory again. In addition, a method of performing an in-frame area detection process by software using a CPU or the like according to the algorithm of the present embodiment is also possible. In this case, the processing speed is extremely high, although the area detection capability may be inferior to an algorithm for detecting an area in a frame by randomly scanning an image.

Next, a color image editing apparatus according to a second embodiment of the present invention will be described. This embodiment is different from the first embodiment in the in-frame area detection processing. Therefore, here, the in-frame area detection processing in this embodiment will be described. FIG. 21 shows an algorithm of the in-frame area detection processing. In this algorithm, an attribute called a ground level is given to the ground pixel, and the ground level of the ground portion of the current raster is determined based on the frame signal of the previous raster, the ground level, the frame signal of the current raster, and the ground level of the current raster already determined. Then, the inside and outside of the frame are determined based on the ground level. This processing is sequentially repeated while advancing one raster at a time, thereby detecting an in-frame area for the entire image.

FIG. 22 shows a procedure for determining the ground level of the ground portion of the current raster. First, the 0th raster is virtually considered as in the first embodiment. In the 0th raster, all pixels are ground and the ground level is 0. Next, the ground level of each raster after the first raster is determined in the following procedure.

First, the frame signal P (x) of the raster is converted into a run expression. Then, the ground level is determined for the ground run according to the following rules.

If there is only one run on the ground before the run Rn, the ground level of the run is set to the ground level of the run Rn [FIG. 23 (a)].

When there are a plurality of runs on the ground adjacent to the run Rn in the preceding raster, the minimum value among the ground levels of the plurality of runs is set as the ground level of the run Rn [FIG. 23 (b)].

And a run that does not satisfy these conditions, that is,
If there is no run in the previous raster adjacent to Rn [No.
23 (c)] shows that, among the runs whose levels have been determined by the above or rules, the two run levels sandwiching the run Rn and closest to the run Rn are L1 and L2, and the following relationship is obtained based on the relationship between L1 and L2. The rule determines the level Ln of the run Rn. However, when there is no run whose level has been determined on both sides of the run Rn or any of them, the level on the side having no determined run is determined as 0.

If L1-L2 is odd, Ln = max (L1, L2) If L1-L2 is even, Ln = max (L1, L2) +1 According to the rule, the ground part connected in the main scanning direction and sub-scanning direction is Inherit other levels. In the case where a new ground portion is separated from the other portion by an upwardly convex frame line (FIG. 23 (a)), a value obtained by adding 1 to the level of the outer ground portion is newly generated by a rule. The ground level is the ground level, and the ground level inside the frame is larger than the ground level by one. Therefore, as in the first embodiment, the ground level determined in this way represents an out-of-frame area if an even number, and an in-frame area if an odd number. In addition, even if the frame has a beard, rules other than the beard are correctly determined according to the rules. Therefore, by performing the same processing as in the first embodiment, the in-frame area can be correctly detected even with respect to a frame of a double or more frame or a frame of a downward concave portion.

[Effects of the Invention] As described above in detail, by performing the in-frame detection using the present invention, the lower side can be obtained even in an area surrounded by a double or more frame line or a frame line having a concave portion formed below. The inside of the concave portion or the double frame can be detected as being outside the frame, and a reasonable in-frame detection can be performed topologically and with common sense. In addition, stable detection can be performed even on a beard on a frame generated due to handwriting. In addition, by performing thinning with priority given to the frame portion before determining the level, the signal amount can be reduced without losing the frame line. In addition, when the level is determined, the processing is performed in units of runs, so that the level can be determined by one scan. By using these processes together, the scale of the in-frame detection process can be simplified, and the present invention can be realized efficiently.

[Brief description of the drawings]

1 to 20 are views for explaining a color image editing apparatus according to a first embodiment of the present invention. FIG. 1 is a block diagram showing the overall configuration of the apparatus, and FIG. , FIG. 3 is a diagram for explaining the operation of the thinning processing unit, FIGS. 4 and 5 are flowcharts showing an algorithm of the in-frame area detecting unit, and FIG. FIG. 7 is a diagram for explaining the connection relationship between two adjacent rasters, FIG. 7 is a diagram for explaining the adjacency relationship of the pixels, FIG. 8 is a diagram for explaining the level determination within the frame, FIG. Is a diagram for explaining an example of detection of the in-frame region of the circular frame line, FIG. 10 is a diagram for explaining an example of detection of the in-frame region of the frame line having the lower concave portion, and FIG. 11 is a donut shape FIG. 12 is a diagram showing a region, FIG. 12 is a diagram showing a frame for designating a donut-shaped region, and FIG. 13 is detection of a region within a frame of a triple frame. FIG. 14 is a diagram illustrating an open frame, FIG. 15 is a diagram illustrating a bearded frame, and FIG. 16 is a diagram illustrating a specific configuration of an in-frame area detection unit. ,
FIG. 17 is a diagram showing the input / output relationship of the level determination logic circuit in the in-frame region detection unit, FIG. 18 is a diagram showing the input / output relationship of the encoder circuit in the inside region detection unit, and FIG. FIG. 20 is a diagram showing an output relationship, FIG. 20 is a diagram showing an example of an image editing process, and FIGS. 21 to 23 are diagrams for explaining a color image editing apparatus according to the second embodiment.
FIG. 22 is a flowchart showing an algorithm for detecting an area within a frame, and FIG.
The figure is a diagram for explaining the adjacent two raster connection relationship.
FIG. 24 is a diagram for explaining the problem of the conventional in-frame area detection. 101 image input unit, 102 color determination unit, 103 thinning processing unit, 104 area detection unit within frame, 105 image editing processing unit, 106 image recording unit, 1601 to 1604 Level determination section, 1605 to 1608 Memory section, 1609 Determination logic circuit, 1610 Internal state register, 1611 Output register, 1612, 1613 Memory, 1614, 1615 Address counter, 1616 Read register, 1617 ... Comparator, 16
18 ... Pixel position counter, 1619 ... Switcher, 1620 ...
Logic circuit, 1631 ... encoder, 1641 ... decoder.

──────────────────────────────────────────────────続 き Continuation of the front page (56) References JP-A-57-176482 (JP, A) JP-A-63-56763 (JP, A) JP-A-61-80473 (JP, A) (58) Field (Int.Cl. ⁶ , DB name) G06T 7/40

Claims (2)

(57) [Claims] 1. An identification means for identifying an image signal obtained by a raster scan into a frame portion and a ground portion other than the frame portion, and a connection area of a frame portion of each raster of the image signal identified by the identification means. Storage means for storing respective levels of a frame run and a ground run which is a connection area of the ground portion, and the levels stored by the storage means as the respective levels of the frame run and the ground run of the previous raster; Level determining means for determining the respective levels of the frame run and the ground run of the current raster with reference to the current raster and the respective levels of the frame run and the ground run of the current raster determined by the level determining means. Determining means for determining whether the ground portion of the raster is inside or outside the frame; 2. A means for determining whether each pixel is a frame color or a ground color from an image signal, and the image signals are grouped in units of a certain number of pixels, and all the pixels in the group are ground color. Means for outputting a background color to a frame color, and when there is a frame color of one pixel or more in the group, outputting one of the frame colors by thinning out the frame color. Means for detecting an in-frame area.