JPS63155274A - Zero-cross detection system for image - Google Patents

Abstract

PURPOSE:To perform zero-cross detection of an image at a high speed by pipe line processing by ANDing image data obtained by binarizing a primary differen tiating circuit output of an input image into with image data obtained by detecting the zero of a secondary differentiating circuit output. CONSTITUTION:The input image is differentiated primarily by a primary differentiating circuit 1 in picture element units and its output is supplied to an AND circuit 6 through a threshold value processing circuit 2. Further, the input image is differentiated secondarily by a secondary differentiating circuit 3 in picture element units and the zero point of its output is detected by a zero detecting circuit 4 and the output is supplied to an AND circuit 6 after being delayed by a delay circuit 5 for synchronizing a flow of the image data after the primary differentiation with a flow of the image data after the secondary differentiation. The AND circuit 6 ANDs the output image of the threshold value processing circuit 2 with the output image of the zero detecting circuit 4 to detect the zero crossing, i.e. contour part.

Description

[Detailed Description of the Invention] [Summary] In a method for detecting zero crossings in an image, there is a property that in a portion of the original image where there is no change in density, the -order differential value as well as the second-order differential value becomes °0゛. Focusing on By providing a zero detection circuit for the output of the differentiation circuit, the image data obtained by binarizing the output of the first-order differentiation circuit of the input image and the image data in which zero is detected for the output of the second-order differentiation circuit are ANDed. , the zero crossings of the input image are detected by pipeline processing.

[Industrial application field]

The present invention relates to an image processing apparatus, and more particularly to a method for detecting zero crossings in an input image at high speed by pipeline processing.

With recent advances in image processing-related technology, processing of images captured by television cameras has become popular.For example, in automatic speed monitoring devices for automobiles, objects are It is required to perform contour recognition processing at high speed.

In addition, in the field of factory automation (FA), which is being put into practical use as the performance of computer systems improves and becomes more widespread, television cameras, etc. There is a need for an image zero-cross detection method that can quickly recognize the contours of an input image captured by a user.

[Prior art and problems to be solved by the invention] FIG.
) shows an example of the configuration of a second-order differential circuit, (b) shows an example of the overall configuration, (c) shows an example of the configuration of a zero-cross detection method using firmware, and (d)-1 to (d)4 show the corresponding examples. A diagram illustrating the concept of a zero-crossing detection method using firmware is shown.

Conventional image zero-cross detection methods include methods using software and methods using a combination of software and hardware.

In the software method, a processor (MPU) sequentially detects zero crossings of the image data pixel by pixel using a processor (MPU), but this method requires address calculation to refer to the image data in the memory. There was a problem in that the amount of data required was enormous and speeding up could not be achieved.

In the method using both software and hardware, the above address calculation can be avoided, but each pixel in the window set in the image developed on the memory 11 is processed by the processor (MPU).
) 10 requires sequential processing and is not suitable for pipeline processing.

FIG. 4 explains the zero-cross detection method using the above-mentioned software and hardware combination.

This method includes (b), (c), (d)-1 to (d)
- As is clear from Figure 3, human image data is processed in a locally parallel manner using a two-dimensional spatial filter-type second-order differential circuit 3, and a second-order differential value with respect to the boundary of the image is determined by pipeline operation. The differential value is once developed in the memory 11, and the processor (MPtl)
) 10, boundary detection and zero-cross determination are performed.

The above-mentioned two-dimensional spatial filter-type second-order differential circuit 3 has, for example, the configuration shown in FIG. When each pixel (one pixel) is sequentially input to a shift register 31.32, a well-known two-dimensional spatial filter is Based on the principle of , every time a new pixel is input to the second-order differential circuit, the second-order differential value for the middle pixel of the 3×3 matrix is calculated, and ? It is stored in the memory 11 via the IPU bus 12.

In this second-order differential circuit 3, each time a new pixel is input, the second-order differential value for the middle pixel of the 3×3 pixels is calculated for a two-dimensional window consisting of 3×3 pixels of the human image. Therefore, three lines of input image data are processed in parallel to obtain a two-dimensional second-order differential, which is called a locally parallel method.

The principle of the two-dimensional spatial filter is as follows:
By performing a logical product between the fixed value defined as the second-order differential window of ), that is, the point of inflection of the image data is determined.

Next, a conventional image zero-crossing detection method using the second-order differential value by the above hardware will be explained with reference to FIGS. (b), (c), and (d)-1 to (d)-3.

In the conventional method, firmware (see figure (C)) executed by the processor (MPU) 10 determines zero crossing based on the results of boundary detection (1) and (2).

In the boundary detection (1) process, as shown in Figure (d)-1, all the points where the second-order differential value of ゛O゛ or less (including 0゛) and the positive second-order differential value are in contact are detected. A pixel (indicated by an O mark) is detected.

Detection of this boundary pixel is performed by re-extracting the pixel value of a 3×3 window portion, for example, for the second-order differential value of each pixel developed on the memory 11, and searching the window clockwise. For example, by repeatedly finding a pixel that changes from positive to negative and moving the window to that point, the second differential point (i.e., the eccentric point of the input image)
Detect.

Similarly, in the boundary detection (2) process, (d) -
As shown in Figure 2, all pixels (O
) is detected.

In the zero cross determination, as shown in Figure (d)-3, the pixels (
) is determined, and the pixel is determined as the zero-crossing point of the input image.

In this way, the conventional zero-crossing detection method using a combination of hardware and software requires random access to the memory 11 in boundary detection (1) and (2) processing by firmware, so the above-mentioned secondary There is a problem in that the processing in the differentiating circuit 3 cannot be pipelined, and as a result, the speed of zero-cross detection cannot be increased.

In view of the above-mentioned conventional drawbacks, the present invention provides that, in areas where there is no density change in the original image, the -th differential value is
By focusing on the fact that there is a property that the density change becomes 0'', we detect only the bending part of the density change in the contour part of the original image (i.e., the pixel where the second-order differential value becomes zero), and detect the zero cross of the image at high speed. The purpose is to provide a method to do so.

[Means for solving problems]

FIG. 1 is a diagram showing an example of the configuration of an image zero-cross detection method according to the present invention.

The present invention is a method for detecting zero crossings in an image, and includes a first-order differentiation circuit 1 that performs first-order differentiation of an input image pixel by pixel and detects its size, and binarizes the output of the first-order differentiation circuit 1. a second-order differentiation circuit 3 for detecting the magnitude of the second-order differentiation of the input image pixel by pixel; a circuit 4 for detecting zero in the output of the second-order differentiation circuit 3; It is equipped with a delay circuit 5 for synchronizing the subsequent flow of image data and the flow of image data after second-order differentiation, and the threshold processing result after first-order differentiation and the zero of the image data after second-order differentiation are provided. The configuration is configured to perform a logical product with the detected result to detect zero crossings in the image.

[Effect]

That is, according to the present invention, in a method for detecting zero crossings of an image, in a portion of the original image where there is no density change,
Focusing on the fact that there is a property that the -th differential value becomes 0゛ along with the second-order differential value, a first-order differential circuit in the form of a two-dimensional spatial filter that can process an input image in a locally parallel manner, for example,
By providing a second-order differentiator, a circuit that binarizes the output of the first-order differentiator, and a zero detection circuit for the output of the second-order differentiator, image data obtained by binarizing the output of the first-order differentiator of the input image is obtained. The zero crossing of the input image is detected by pipeline processing by performing a logical AND with the image data in which zero is detected for the output of the second-order differential circuit. This has the effect of enabling high-speed processing.

〔Example〕

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

The above-mentioned FIG. 1 is a diagram showing a configuration example of the image zero-cross detection method of the present invention, FIG. 2 is a diagram explaining the zero-cross detection operation according to the present invention, and FIG. 3 is a diagram showing the logic of the differential circuit. FIG. 2 is a diagram illustrating the first-order differential circuit 1 in FIG. Threshold processing circuit 2° Zero detection circuit 4. Logical product (AN
D) Circuit 6 is the necessary means to implement the invention. Note that the same reference numerals indicate the same objects throughout the figures.

Hereinafter, the zero cross detection method of the present invention will be explained with reference to FIGS. 2 and 3 while referring to FIG. 1.

First, the logic of the differential circuit will be explained with reference to FIG.

Differentiation for image data is achieved, for example, by processing based on the principle of a known two-dimensional spatial filter.

The principle of the two-dimensional spatial filter is that for pixel P(i,j), however,! =Rin=0. Based on the formula shown as ±1, Figure 3 (a), (b),
For example, the 3×3 window °a1m shown in (c)
A logical AND operation is performed between the fixed value of ' and the pixel value obtained by scanning the original image in the same 3 x 3 window, and the summation °0(IIJ)' is calculated as the pixel P(i, j ).

(a) in this figure is the window °a for obtaining the X-direction differential value.
.

The window “al” for obtaining the X-direction differential value in (a) above
To explain this in terms of ``va'', if the shading of the image in the When calculated, it is 0'', but if there is a shading between the pixels, it can be seen that the summation shows a specific value, that is, a differential value in the X direction, depending on the degree of the shading.

Therefore, by performing such a calculation every time each pixel is input to the differentiating circuit, the differentiating process is performed in a Vibrine manner, and the differential value corresponding to each pixel P (i, j) is obtained. Output.

The present invention attempts to detect zero crossings in an image using a Vibrine method using a differential circuit based on the principle of a two-dimensional spatial filter.

As shown in Fig. 3, the primary differentiation circuit 1 in Fig. 1 scans the original image in a 3 x 3 window, calculates the differential values in the By calculating the square root, the magnitude of the first-order differential value of the original image is calculated. (See Fig. 2(c)) The next threshold processing circuit 2 applies a threshold value (±th) to the differential image shown in Fig. 2(c), and as shown in Fig. 2(d), A portion where the differential value is large is detected.

The second-order differential circuit 3 calculates the second-order differential value of the original image using 3×3 fixed mask data as shown in FIG. 3(c). (See FIG. 2(e)) The zero detection circuit 4 detects pixels whose pixel value is 0° in the image data consisting of the second-order differential values shown in FIG. 2(e). (See FIG. 2 Cf)) A logical product (AND) circuit 6 calculates the logical product of the output image of the threshold processing circuit 2 and the output image of the zero detection circuit 4 for each corresponding pixel. (See Figure 2 (g)) In areas with no shading in the original image, both the -th differential value and the second-order differential value are 0゛, so the pixel for which "1°" was obtained in the above logical product is the -th order differential value. This means that the differential value is 1' and the secondary differential value is 0, which means that the zero cross of the original image (see FIG. 2(a)), that is, the contour portion has been detected.

As mentioned above, the first-order differential circuit 1. Since the second-order differential circuit 3 is pipeline-processed using a local parallel processing method, the zero-crossings of the original image can be pipeline-processed by detecting the zero-crossings of the image using the zero-crossing detection circuit shown in FIG. It is possible to detect the contour part in the grayscale image of the original image at high speed.

Note that the delay circuit shown in the configuration example of FIG. 1 is a first-order differentiator circuit 1. and a pipeline delay composed of a threshold value processing circuit 2, and a second-order differentiation circuit 3. and the vibration line delay of the zero detection circuit 4 is corrected.

As described above, the present invention focuses on the characteristic that in a part of the original image where there is no density change, the second-order differential value becomes °O'' along with the -th-order differential value, and the second-order differential value of the original image is For the pixels for which zero is detected, the -th differential value is ANDed with the pixel value sliced by a certain darkness value, thereby extracting pixels only in the curved part of the density change in the contour part of the original image, The feature is that the calculation part of the -th order differential and the second order differential is constructed using a locally parallel method, and the zero cross detection of the image is performed by pipeline processing.

〔Effect of the invention〕

As explained above in detail, the image zero-crossing detection method of the present invention detects the zero-crossing of an image, and in the portion of the original image where there is no density change, the -th differential value as well as the second-order differential value is detected. Focusing on the property that the input image becomes 0", we have developed a two-dimensional spatial filter-type first-order differentiator circuit, a second-order differentiator circuit, and an output of the first-order differentiator circuit that can process an input image, for example, in a locally parallel manner. By providing a circuit for binarizing the output of the first-order differentiator and a zero detection circuit for the output of the second-order differentiator, it is possible to detect zeros for the image data obtained by binarizing the output of the first-order differentiator of the input image and the output of the second-order differentiator. Since the zero cross of the input image is detected by pipeline processing by performing a logical product with the detected image data, the effect is that the zero cross detection of the image can be processed at high speed by pipeline processing. be.

[Brief explanation of the drawing]

FIG. 1 is a diagram showing an example of the configuration of an image zero-cross detection method according to the present invention. FIG. 2 is a diagram illustrating the zero-cross detection operation according to the present invention. FIG. 3 is a diagram explaining the logic of the differential circuit. FIG. 4 is a diagram illustrating a conventional image zero-cross detection method. It is. In the drawings, 1 is a first-order differential circuit, and e is a second-order differential circuit. You are the threshold processing circuit, and 4 is the zero detection circuit. 5 is a delay circuit, and 6 is an AND circuit. are shown respectively. ! ? Xar I1, P(ld) (0) X direction throw win, 1 lamp fixed heat (Jitsu (i, j) = Σ Σ2 θt7'Ipα+l, jfyn-) 1
i a note r) famous accountant left shoulder figure I saw the other day - figure 3 on the wall

Claims (1)

[Claims] A method for detecting zero crossings in an image, comprising: a first-order differentiation circuit (1) that performs first-order differentiation of an input image pixel by pixel and detects its size; a threshold processing circuit (2) that binarizes the output; a second-order differentiation circuit (3) that detects the magnitude of the second-order differentiation of the input image pixel by pixel; and an output of the second-order differentiation circuit (3). Circuit to detect zero (4
), and a delay circuit (5) for synchronizing the flow of image data after the first differentiation and the flow of image data after the second differentiation, and the threshold processing result after the first differentiation and the second An image zero-cross detection method characterized by detecting a zero-cross in an image by performing an AND with a result of detecting zero in image data after differentiation.