JPH0423303B2 - - Google Patents

Description

[Detailed Description of the Invention] (1) Technical Field of the Invention The present invention compresses two-dimensional information of an input image in the form of a segment table, and stores information in the scanning line direction in a buffer memory in the form of a run (a series of pixels). The present invention relates to a method in which the correlation between scanning lines is determined by comparing the run information with a segment table, and the coordinate values of a vector (broken line approximation pattern) are promptly output when vector generation conditions are satisfied. .

(2) Conventional technology and problems Conventionally, image processing equipment simply photoelectrically converts documents and drawings and performs only binary processing (converting shading into two values of white and black). The drawbacks were that the storage capacity of the converted data was large and that it could not be easily used for pattern processing.

(3) Purpose of the Invention The present invention solves the above-mentioned drawbacks, photoelectrically converts line figures, further performs binarization processing, and then quickly obtains vectorized (broken line approximation) data, thereby converting image data. The purpose is to facilitate encoding and pattern recognition processing.

(4) Structure of the invention The object is to provide a system that photoelectrically converts a line image and binarizes it into 0 or 1 using a binarizer, which includes a buffer memory for storing the binarized data, and a buffer memory for storing the binarized data;
a run detection means for detecting a run from digitized data;
The number of runs in the detected row and the number of runs in the previous row in the section corresponding to the runs in the detected row are counted, and the number of runs in the detected row and the run in the previous row are counted. a matching means for comparing the number of runs, a segment table in which the segment length, slope, and line width of a segment composed of a collection of runs are registered; If the number of runs to be combined is one, the segment length, slope, and line width of any segment registered in the segment table are compared with the run, and the positions of the pixels at both ends of the run are registered. By checking whether the run is apart from the slope of the segment or whether the run width deviates from the registered line width of the segment, it is possible to determine whether the run is included in the segment or is a bend. a segment determination means for determining and registering the run in the segment table, and registering the run as a new segment in the segment table if the number of runs to be combined is zero or more; and the highest segment among the registered segments. and vector generation means for generating a vector from the coordinate relationship of the lowest segment.

(5) Embodiments of the invention The present invention will be described in detail below with reference to the drawings.

FIG. 1 is an overall block diagram showing one embodiment of the present invention.

In the figure, 11 is a drawing, 12 is a photoelectric conversion unit,
13 is a binarizer, 14 is a row counter, 15 is a column counter, 16 is a data bus, 17 is a switch, 1
8 is a buffer memory A, 19 is a memory access unit 1 (memory read/write), 110 is a run detection unit,
111 is a memory access unit 4, 112 is a buffer memory B, 113 is a memory access 2, 114 is a matching unit, 115 is a segment table, 1
16 is a memory access 3, 117 is a segment determination section, 118 is a vector generation section, and 119 is an external memory.

FIG. 2 is a diagram showing details of run detection according to an embodiment of the present invention.

In the figure, 21 is l ₁ (segment label), 2
2 is black part, 23 is l ₂ (segment label), 24
is the black part, 25 is the buffer of the previous row, and 26 is the buffer of the processed row.

FIG. 3 is a diagram showing details of a segment table showing an embodiment of the present invention. In the figure, 31
is the segment number, 32 is the segment start point coordinates,
33 is the segment length, 34 is the slope, and 35 is the line width.

First, from left to right in drawing 11 with photoelectric conversion unit 12,
Television-style scanning is performed horizontally one line at a time from top to bottom, and the binarizer 13 converts the data into 0 (white) and 1 (black).
The data is converted into a value and sequentially stored one row at a time in the buffer memory A 18 and buffer memory B 112 using the row counter 114. The buffer memory has n bits per pixel because it can handle integer values. The width of the buffer and the number of bits n can be changed depending on the image being handled.

Next, the binary data stored in the buffer memories A18 and B112 is transferred to the memory access units 19 and 1.
13, and the run detection unit 110 reads 1.
One series of (black) parts is detected as a run (RUN). That is, the binary data stored in the buffer memory is detected from one direction, and the part where the value is greater than 0 (the start of the run) and the opposite part (the end of the run) are recorded. The matching unit 114 examines the combination of the runs, and as shown in FIG. 2, it examines the runs of the buffer 25 in the previous row and the buffer 26 in the processing row. In other words, the beginning
In the case of a run whose end coordinates are x ₁ and x ₂ , the section of the previous buffer 25 corresponding to x ₁ to _{x 2} is stored in column counter 1.
5 to see if there is a run with a value greater than 0 (l ₁ 21). We are counting how many such runs there are in the interval [x ₁ , x ₂ ]. If it is 0, it means that there is no run connected to the previous row and that the run appears for the first time in the processing row buffer 26. K pieces (K>
0) then K runs (segments or lines)
This shows that the lines have gathered into one line. When the above processing is completed for one run, the next segment determination unit checks the run connection state to determine which segment (run collection) the run currently being processed is connected to. Here, a segment is a collection of runs, and corresponds to a line segment with a certain inclination. In this explanation, the previous buffer 25
The value l ₁ 21 found when scanning is the segment number. Segment length 33, i.e. 1 for the first run
and +1 value if it extends downward, slope 34 indicating the average slope between the first run and the next run, and line width 35 which is the difference in run coordinates between the first run and the next line.
A segment table 115 with the following information is created.
Subsequently, the next run is written in the segment table 115 in the same manner. Segment table 11
5, the segment determination unit then compares the average slope Δx, line width W, etc. of the segment l ₁ 21 with the coordinates of x ₁ and x ₂ of the current run, and determines the current run. l If x _{1 and} x ₂ deviate greatly from Δx or the line widths are extremely different, do not include _them in the same segment; If x ₂ does not deviate much and the line widths are approximately the same, it is determined to be included. At that time, if it is determined that they are included in the same segment, m is incremented by 1 and l ₁ is written in the run part of the buffer. On the other hand, if it is determined that the run is not included in the same segment, a new number _l1 is given to the run and the segment table 11 is
Register for 5. Next, after all of the runs for one row are registered in the segment table 115 as segments, the vector generation unit 118 vectorizes the segments if the following conditions are satisfied.

′ If a new segment occurs below the segment (if a run not included in segment l ₁ creates a new segment l ₁ ...corresponds to a bend in the line) (This corresponds to an interruption.) Vectorization of the segment is performed from the starting point coordinates x, y, slope Δx, and length m in the segment table 115. (Xf, Yf) ← (x, y):
Yt)←(x+m・Δx,y+m): Calculate the vector [(Xf, Yf)→(Xt, Yt)], where t is an abbreviation for to.

After the above processing for one row is completed, the buffer under processing is transferred to the previous row buffer 25, the binary data of the next row is input into the buffer under processing, and a vector is generated in the same manner.

(6) Effects of the Invention As described above, according to the present invention, binarized data can be quickly vectorized (broken line approximation) and pattern recognition processing can be facilitated.

[Brief explanation of drawings]

FIG. 1 is an overall block diagram showing one embodiment of the present invention. FIG. 2 is a diagram showing details of run detection according to an embodiment of the present invention. FIG. 3 is a diagram showing details of a segment table showing an embodiment of the present invention. Explanation of symbols, 11 is a drawing, 12 is a photoelectric conversion unit,
13 is a binarizer, 14 is a row counter, 15 is a column counter, 16 is a data bus, 17 is a switch, 1
8 is a buffer memory A, 19 is a memory access unit 1 (memory read/write), 110 is a run detection unit,
111 is a memory access unit 4, 112 is a buffer memory B, 113 is a memory access 2, 114 is a matching unit, 115 is a segment table, 1
16 is a memory access 3, 117 is a segment determination section, 118 is a vector generation section, and 119 is an external memory.

Claims (1)

[Claims] A system that photoelectrically converts a one-line image and binarizes it into 0 or 1 using a binarizer, comprising: a buffer memory for storing the binarized data; and a run from the binarized data. A run detecting means for detecting, counting the number of runs in the detected row and the number of runs in the previous row of the section corresponding to the run in the detected row, and counting the number of runs in the detected row. a matching means that compares the number of runs with the number of runs in the previous row; a segment table in which the segment length, slope, and line width of a segment composed of a collection of runs are registered; and an arbitrary run from the number of runs. If the number of runs to be combined is one, compare the segment length, slope, and line width of any segment registered in the segment table with that run, and compare both ends of the run. Whether the run is included in the segment by checking whether the position of the pixel is not far from the slope of the registered segment or whether the run width does not deviate from the line width of the registered segment; or a bent part, and registers the determined run in the segment table, and if the number of runs to be combined is zero or more, registers the run as a new segment in the segment table; A vector generation method for a line image, comprising: vector generation means for generating a vector from the coordinate relationship between the highest segment and the lowest segment among the segments;