JPS6410871B2 - - Google Patents

Description

[Detailed Description of the Invention] [Technical Field of the Invention] The present invention performs a spatial filtering operation on an image signal of an object extracted by an imaging device such as a television camera to enhance contours and remove noise components. The present invention relates to an image signal processing method and apparatus for performing.

[Prior art]

As an image processing technique that has recently become frequently used, there is a technique called spatial filtering operation. This converts the brightness information of a two-dimensional input image into 2
This is an operation for obtaining a new image through dimensional processing, and among these spatial filtering operations, a process called "convolution wave" is particularly frequently used.

FIG. 1 is a diagram for explaining this convolutional wave processing, in which the number of pixels of the image obtained by the imaging device is ``x×y'', and the coordinates of each pixel are 1,
1, 1, 2, ... x, y, each pixel is the first
As shown in figure a, a, 1, 1, a, 1, 2,...
Represented as a, x, y.

Pixels a, m,
n [m=1~x, n=1~y] and each surrounding pixel, p elements are arranged in the horizontal direction and q elements in the vertical direction, for example, an operator b as shown in FIG. 1b. ,1,1,b1,2,...b,3,3, c(m,n)=b(1,1)・a(m-1,n-1)+b(1,2 )・a(m-1,n) +b(1,3)・a(m-1,n+1) +b(2,1)・a(m,n-1) +b(2,2)・a(m ,n)+b(2,3)・a(m,n+1)+b(3,1)・a(m+1,n-1)+b(3,2)・a(m+1,n)+b(3, 3)・a(m+1,n+1)...(1) A converted pixel c as shown in FIG. 1c,
m and n are obtained.

In order to sequentially perform the operation of determining the converted pixels c, m, n for one central pixel m, n for all the pixels shown in FIG.
While moving the operator shown in pixel by pixel,
Each time the image is moved, the calculation shown in equation (1) above is executed, and the transformed image obtained in this way has a spatial frequency different from that of the original image.

For example, an operator consisting of a 3x3 matrix is In this case, this operator functions as a low-pass filter, so images with low spatial frequencies are emphasized. Also the operator In this case, this operator functions as a high-pass filter, so images with high spatial frequencies are emphasized.

By the way, in order to realize such a filtering operation, the image signal obtained from the imaging device is converted into a digital signal, stored in a frame memory, and sequentially accessed and calculated by a calculation device and output. It is necessary to repeat this operation for all pixels in one frame.

However, calculations for such filtering operations involve multiplication processing over multiple terms, and this operation must be performed for each pixel, which requires a long processing time and is not suitable for research purposes. There is a problem that it cannot be put to practical use in any other way. For example, when using an operator consisting of a 3×3 matrix, in order to obtain converted pixels c, m, and n for one pixel, it is necessary to access the frame memory nine times and perform a product-sum operation over nine terms.
Therefore, if one frame consists of 512 x 512 pixels, in order to process one frame,
Requires 2359296 accesses (= 512 x 512 x 9) and sum-of-products operations, with one access time of 1 μsec
Then, the total access time for one frame is approximately 2.4 seconds (=2359296×1×10 ^-6 [sec]), and if the time for one product-sum calculation is 10 μsec, the total product-sum calculation time for one frame is approximately 24 seconds. seconds (=2359296×10×10 ^-6
[sec]). When the size of the operator is large, the processing time also increases in proportion to the square of the size.

Traditionally used spatial filtering operations require long processing times;
The problem is that it is very impractical for purposes other than research.

[Object and structure of the invention]

The present invention has been made in view of the above circumstances, and its purpose is to provide an image signal processing method and apparatus that can remove noise components of an object image and enhance contours in a short period of time. It is in.

To this end, the present invention sets p pixel regions in the horizontal direction and q pixels in the vertical direction as an operation unit, and then repeatedly extracts the object image signal over a plurality of image frames. Weights given to each image signal within an operation unit while shifting the relative positional relationship of pixels between each image frame among the signals by one pixel in the horizontal and vertical directions for each frame corresponding to p×q operation units. The image signals of the number of frames corresponding to the coefficients are subjected to addition/subtraction operations, and the results of the operations are output as converted image signals. That is, instead of multiplying each image signal within an operation unit by a weighting coefficient, each image signal is added or subtracted a number of times corresponding to the weighting coefficient, and the image signals surrounding the center pixel are When adding or subtracting, the count value of the address counter that specifies the address of the frame memory is offset to an address based on the center pixel, and the pixel positional relationship between the frames is determined relative to each other by starting from this offset address. Furthermore, this address offset is performed every time one frame of image extraction is completed.
This makes it possible to complete the product-sum calculation within an operation unit in the image extraction time of the number of frames equal to the total sum of weighting coefficients.

〔Example〕

FIG. 2 is a block diagram showing an embodiment of an apparatus to which the image signal processing method according to the present invention is applied, and FIG. 3 is a diagram for explaining the image processing method. In FIG. 2, 1 scans an object (not shown) and
an imaging device for extracting an analog image signal AV of an object consisting of, for example, 512 x 512 pixels per frame; 2;
is an analog image signal output from the imaging device 1
An AD converter that converts AV into a digital image signal V, and 3 an arithmetic unit that adds or subtracts the digital image signal V output from the AD converter 2 and a reproduced digital image signal RV read out from a frame memory 4, which will be described later. , 4 is a frame memory for storing the digital image signal V' output from the arithmetic unit 3, and has a storage capacity sufficient to store one frame of image signal. 5 is the reproduced digital image signal RV read out from the frame memory 4
A DA converter that converts the image into an analog image signal ATV,
6 is a display device that receives the analog image signal ATV and displays it as a visible image; 7 is a display device that distributes and supplies necessary synchronization signals to each of the units 1 to 6 described above, and also sets the size and weighting coefficient of the operating unit (operator); Based on this, an addition or subtraction command signal A/S is given to the arithmetic unit 3, and a cycle command signal R/W/A for reading, writing, and calculation and an address signal ADR are given to the frame memory 4. This control circuit has the function of shifting the contents of the address signal ADR one pixel at a time in the horizontal and vertical directions in accordance with the unit of operation.

In such a configuration, the control circuit 7 provides a synchronization signal to the imaging device 1, and causes the imaging device 1 to repeatedly extract one frame of image signal AV according to the shade of the object. In addition, a synchronization signal is also given to the AD converter 2, and the image signal is repeatedly extracted.
The AV is converted into a digital image signal V, and this digital image signal V is inputted to one calculation input A of the calculation unit 3.

On the other hand, the control circuit 7 provides the frame memory 4 with an address signal ADR that changes in synchronization with the change in the sequential scanning points of the object by the imaging device 1, and also provides the frame memory 4 with an address signal ADR that changes in synchronization with the change in the sequential scanning points of the object by the imaging device 1. ), first frame memory 4
Cycle command signal R/ to set the read mode.
W/A, then a cycle command signal R/W/A that starts the calculation operation of the arithmetic unit 3, and finally a cycle command signal R/W/A that sets the frame memory 4 to write mode. Output. Also,
A command signal A/S is outputted to the arithmetic unit 3 to cause the arithmetic unit 3 to perform addition or subtraction until the arithmetic operation corresponding to a certain weighting coefficient within the unit of operation is completed.

Here, as shown in Fig. 3a, an operation unit of size consisting of _b0 to _b8 is set, and the weighting coefficients k[ _b0 ] to k[ _b8 ] for each pixel are shown in Fig. Assuming that such a value is set, the control circuit 7 first converts the image signal V of the center pixel into its weighting coefficient k [b ₀ ].
= 4, the pixel positional relationship of the digital image signal V and the reproduced digital image signal RV match each other, and the pixel coordinates of the reproduced digital image signal RV are "x=0, y=0". Address signal starting from and ending with "x=512, y=512"
Output ADR repeatedly four times. That is, the control circuit 7 outputs the address signal ADR with zero address offset to the frame memory 4 for four frames in synchronization with the image extraction operation by the imaging device 1. At the same time, since the weighting coefficient k[b ₀ ] is a positive value, the arithmetic unit 3 is instructed to perform an addition operation.

Then, in the arithmetic unit 3, an addition operation between the image signal V outputted from the AD converter 2 and the reproduced digital image signal RV read out from the frame memory 4 is performed for signals related to the same pixel. Since the frame memory 4 is set to the write mode by the signal R/W/A after the calculation is completed, the signal V' resulting from this addition is written to the same address as the read address of the signal RV at this time. Such an operation is executed every time the pixel coordinates indicated by the address signal ADR changes, and is executed until the extraction of four frames of image signals AV is completed.

As a result, at the stage when the extraction of the image signal AV of four frames is completed, each address of the frame memory 4 has the image signal V at each pixel within one frame.
An image signal 4·V obtained by multiplying by 4 is stored.

Next, the control circuit 7 subtracts the image signal V of the pixel b ₁ within the operation unit twice from the image signal 4·V already stored in the frame memory 4, corresponding to its weighting coefficient k[b ₁ ]=-2. Therefore, the pixel positional relationship between the digital image signal V and the reproduced digital image signal RV is
As shown in Figure c, the address signal ADR is shifted by "+1" in the horizontal direction and "-1" in the vertical direction.
is offset, and the address signal ADR for 512×512 pixels starting from this offset address is output twice. In other words, from the AD converter 2 to the arithmetic unit 3
The pixel coordinates of the image signal V input to
0, y=0'' and ends at ``x=511, y=511,'' for example, if the reproduced digital image signal RV with the coordinates ``x=m, y=n'' is applied to one input B of the calculator 3. At the timing when “x=m+1,y
The initial values of the address signal ADR are offset to "x=0-1=511" and "y=0+1=1" so that the image signal V at the coordinates "=n-1" is given to the other input A. Offset address “x=511,y
512 starting from ``=1'' and ending with ``x=510, y=0''
The address signal ADR for ×512 pixels is repeatedly output twice. At the same time, since the weighting coefficient k[b ₁ ] is a negative value, the arithmetic unit 3 is instructed to perform a subtraction operation.

Then, at the timing when the reproduced digital image signal RV=4·V of the coordinates "x=m, y=n" is given, the arithmetic unit 3 calculates "x=m+1, y=n".
Since the image signal V with the coordinates “-1” is given to one input A, the image signal RV with the coordinates “x=m, y=n”
The operation of subtracting the image signal V at the coordinates "x=m+1, y=n-1" shifted by "+1" in the horizontal direction and "-1" in the vertical direction from is performed every time the address signal ADR changes. Then, the signal “RV-V=4・V-V” as a result of this addition becomes “x=
The data is written to the address of the frame memory 4 indicated by "m, y=n". Such operations are executed until extraction of two frames of image signals V is completed.

As a result, when the extraction of image signals for two frames is completed, each address in the frame memory 4 has V[x=m, y=n]=4・V[x=m,y=n]−2・V [x=m+1, y=n-1]
...(2) An image signal V [x=m, y=n] is stored.

Next, the control circuit 7 causes the arithmetic unit 3 to perform similar calculations corresponding to the weighting coefficients k[ _{b2 ]} to k[b8] of the image signals V of the pixels _b2 to _b8 in the operation unit.

As a result, at the stage where the image signal V of the number of frames equal to the sum of the absolute values of the weighting coefficients k[b ₀ ] to k[b ₈ ] is extracted, V[x=m, y=n] = 4·V[x =m, y=n] -2・V[x=m+1,y=n-1] -2・V[x=m+1,y=n] -2・V[x=m+1,y=n+1] -2・V[x=m, y=n+1] −2・V[x=m−1, y=n+1] −2・V[x=m−1, y=n] −2・V[x=m− 1, y=n-1] -2・V[x=m, y=n-1] ...(3) The image signal V[x=m, y=n] shown in be remembered. That is, each image signal V [x=m, y=n] of 512×512 pixels gives the same result as the conventional method (however, m=0 to
511, n=0 to 511) are stored in the frame memory 4.

The converted image signal thus obtained in the frame memory 4 is converted into an analog image signal ATV by the DA converter 5 and displayed as a visible image on the display device 6.

Therefore, by appropriately setting the weighting coefficients k[b ₀ ] to k[b ₈ ], it is possible to remove noise components and display a visible image with enhanced contours. If the image extraction time per frame is, for example, 33 ms, this visible image will take about 0.7 seconds (33 ms).
×10 ^-3 ×20 = 660ms), which can be displayed in an extremely short time.

〔Effect of the invention〕

As described above, the present invention performs a spatial filtering operation using p.times.q operators by a combination of an offset operation of the pixel positional relationship between frames and an addition/subtraction operation.
Therefore, it is possible to remove noise components and enhance contours in a very short time by only adding and subtracting. Therefore, if applied to various image processing such as X-ray image processing, an extremely practical device can be provided.

[Brief explanation of drawings]

FIG. 1 is a diagram for explaining a conventional image signal processing method, FIG. 2 is a block diagram showing an embodiment of the present invention, and FIG. 3 is a diagram for explaining the image signal processing method of the present invention. be. DESCRIPTION OF SYMBOLS 1... Imaging device, 2... AD converter, 3... Arithmetic unit, 4... Frame memory, 5... DA converter, 6... Display device, 7... Control circuit.

Claims (1)

[Claims] 1. P in the horizontal direction with respect to an image signal consisting of a plurality of pixels extracted by scanning an object with an imaging device.
One image frame is an operation in which a pxq pixel area of q pixels in the vertical direction is set as an operation unit, and each image signal within this operation unit is multiplied by a predetermined weighting coefficient and the sum is calculated. In an image signal processing method that is executed over all pixels of the image to emphasize contours and remove noise components, object image signals are repeatedly extracted over multiple image frames, and each image signal is extracted between the repeatedly extracted image signals. An image signal of a number of frames corresponding to a weighting coefficient given to an image signal within the operation unit while shifting the relative positional relationship of pixels between frames by one pixel horizontally and vertically for each frame corresponding to the operation unit. An image signal processing method that performs addition/subtraction operations to enhance contours and remove noise components. 2 P in the horizontal direction with respect to an image signal consisting of multiple pixels extracted by scanning the object with an imaging device.
Setting a pxq pixel area in the vertical direction as a unit of operation, multiplying each image signal in this unit of operation by a predetermined weighting coefficient and calculating the sum is performed in one image frame. An image signal processing device that performs processing on all pixels to enhance contours and remove noise components includes a computing unit that adds and subtracts image signals of multiple image frames; While repeatedly extracting the frames, the relative positional relationship between the pixels of each image frame is applied to the image signal within the operation unit while shifting the pixel relative positional relationship between each image frame by one pixel corresponding to the operation unit between the repeatedly extracted image signals. An image signal processing device comprising: a control device that causes the arithmetic unit to perform an addition/subtraction operation on an image signal of a number of frames corresponding to a weighting coefficient.