JP2004362143A - Edge detection device, component recognition device, edge detection method, and component recognition method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To stably search the edges of components or characters in order to highly precisely recognize those edges. <P>SOLUTION: A set value A is added to a luminance value L[0] of the 0th pixel P[0] to obtain a luminance value=L[0]+A, and this luminance value is compared with an arbitrary luminance value 202. When the sum=L[0]+A is smaller than the maximum luminance value, a pixel position equal to the sum=L[0]+A is defined as the edge point candidate of the pixel P[0]. In the same way, the edge point candidate of the next pixel P[1] is derived, and the edge point candidates of all the pixels are derived up to the final pixel P[max]. The number of the pixel positions defined as the edge point candidates are integrated, and the most numerous pixel positions are defined as the edge points. <P>COPYRIGHT: (C)2005,JPO&NCIPI

Description

[0001]
TECHNICAL FIELD OF THE INVENTION
The present invention relates to an edge detection device and a component recognition device for detecting an edge of a recognition target such as a component or a character, and an edge detection method and a component recognition method.
[0002]
[Prior art]
In a conventional edge detection method, for example, as described in Patent Document 1 below, a differential process is performed to emphasize the edge of a digital image. Hereinafter, a differentiation process, which is an example of a conventional edge detection method, will be described with reference to the drawings. FIG. 4 shows a conventional edge detection method in which adjacent pixels of the image in the processing area 201 of FIG. 2A are differentiated and shown in a two-dimensional graph. First, the image of the processing area 201 shown in FIG. On the other hand, scanning is performed from left to right to calculate the luminance value L [k] of each pixel P [k].
[0003]
Next, the difference ΔL [k] between the calculated luminance value L [k] of each pixel P [k] and the luminance value L [k + 1] of the right pixel P [k + 1] adjacent to the pixel P [k] = L [k + 1] −L [k] is calculated, and the result ΔL [k] is set as a differential value. A pixel P [k] having a large differential value ΔL [k] has a large luminance difference from an adjacent pixel P [k + 1], and detects a peak of the differential value. The coordinates of the pixel of the detected peak are set as an edge point 401.
[0004]
[Patent Document 1]
JP-A-4-178544 [0005]
[Problems to be solved by the invention]
However, with the above configuration, when a plurality of peaks appear, it is unclear which peak is selected as an edge point, and an accurate edge cannot be obtained. There was a problem that it was not possible. In particular, when illuminating and capturing an image of a component protruding from the substrate, an accurate edge cannot be obtained due to the shadow of the edge of the component.
[0006]
The present invention solves the above problems, and provides an edge detecting device and a component recognizing device, an edge detecting method and a component recognizing method capable of stably obtaining an edge of a component or a character and accurately recognizing the edge. The purpose is.
[0007]
[Means for Solving the Problems]
The present invention, in order to achieve the above object, adding means for adding a predetermined value and a pixel value at each pixel position of a plurality of pixels arranged in a direction intersecting the edge of the recognition object,
The addition value at each of the pixel positions added by the addition means is compared with the pixel value at each of the pixel positions of a plurality of pixels arranged in a direction intersecting with the edge of the object to be recognized, and a pixel that substantially matches is determined as an edge candidate. Edge candidate pixel calculating means for calculating as a pixel,
Edge pixel determining means for integrating the number of edge candidate pixels calculated by the edge candidate pixel calculating means for each pixel position, and determining a pixel having the largest number of edge candidate pixels as an edge pixel,
Configuration.
[0008]
Further, according to the present invention, an imaging means for imaging the component in a direction intersecting the edge of the component,
A brightness value calculation unit that calculates a brightness value of each pixel position of a plurality of pixels arrayed in the intersecting direction captured by the imaging unit;
Adding means for adding a predetermined value and a luminance value at each pixel position calculated by the luminance value calculating means,
An edge candidate pixel for comparing an addition value at each of the pixel positions added by the addition means with a brightness value at each of the pixel positions calculated by the brightness value calculation means to calculate a pixel that substantially matches as an edge candidate pixel Calculating means;
Edge pixel determining means for integrating the number of the edge candidate pixels calculated by the edge candidate pixel calculating means for each of the pixel positions, and determining a pixel having the largest number of the edge candidate pixels as an edge of the component. ,
Configuration.
[0009]
With the above configuration, edges of components, characters, and the like can be obtained stably and can be accurately recognized.
[0010]
Further, the component recognition device of the present invention further calculates an edge candidate pixel in the direction opposite to the direction intersecting with the edge by the edge detection device having the above configuration, and calculates the edge candidate pixel in the direction intersecting with the edge and in the direction opposite thereto. The number is integrated for each pixel position, the pixel having the largest number of edge candidate pixels is determined as an edge pixel, and the center surrounded by the edge pixel is determined as the center of the component.
[0011]
With the above configuration, the edge of the component can be obtained stably and recognized with high accuracy.
[0012]
Further, the edge detection method of the present invention, an addition step of adding a predetermined value and a pixel value at each pixel position of a plurality of pixels arranged in a direction intersecting the edge of the recognition target object,
The addition value at each of the pixel positions added in the adding step is compared with the pixel value at each of the pixel positions of a plurality of pixels arranged in a direction intersecting with the edge of the recognition target, and a pixel that substantially matches is determined as an edge candidate. An edge candidate pixel calculating step of calculating as a pixel;
Edge pixel determining step of integrating the number of the edge candidate pixels calculated in the edge candidate pixel calculating step for each pixel position, and determining a pixel having the largest number of the edge candidate pixels as an edge pixel,
Configuration.
[0013]
Further, the edge detection method of the present invention, an imaging step of imaging the component in a direction intersecting the edge of the component,
A brightness value calculation step of calculating a brightness value of each pixel position of a plurality of pixels arranged in the intersecting direction captured by the imaging step;
An adding step of adding a predetermined value and a luminance value at each of the pixel positions calculated by the luminance value calculating step;
An edge candidate pixel for comparing an addition value at each of the pixel positions added in the addition step with a brightness value at each of the pixel positions calculated in the brightness value calculation step to calculate a pixel that substantially matches as an edge candidate pixel A calculating step;
An edge pixel determining step of integrating the number of the edge candidate pixels calculated in the edge candidate pixel calculating step for each of the pixel positions, and determining a pixel having the largest number of the edge candidate pixels as an edge of the component. ,
Configuration.
[0014]
With the above configuration, edges of components, characters, and the like can be obtained stably and can be accurately recognized.
[0015]
Further, in the component recognition method according to the present invention, the edge detection method according to claim 5 further calculates an edge candidate pixel in a direction opposite to the direction intersecting with the edge, and calculates a direction intersecting with the edge and an edge in a direction opposite thereto. The number of candidate pixels is integrated for each pixel position, the pixel having the largest number of edge candidate pixels is determined as an edge pixel, and the center surrounded by the edge pixel is determined as the center of the component.
[0016]
With the above configuration, the edge of the component can be obtained stably and recognized with high accuracy.
[0017]
BEST MODE FOR CARRYING OUT THE INVENTION
Hereinafter, embodiments of the present invention will be described with reference to the drawings. FIG. 1 is a block diagram showing an embodiment of an edge detecting device and a component recognizing device according to the present invention, FIG. 2 is a graph showing a luminance value near the edge of a component, and FIG. FIG. 4 and FIG. 4 are explanatory diagrams showing a conventional edge detection method.
[0018]
In FIG. 1, first, the image input unit 101 captures an image of a planar image or a three-dimensional object whose surface is not on the same plane as the background by using a camera (not shown). In the subsequent processing area setting unit 102, FIG. 2A shows a processing area 201 for detecting an edge of a planar image captured by the image input unit 101 or an edge of a three-dimensional object whose object surface is not on the same plane as the background. Set as follows. In this processing area 201, a plurality of pixels are arranged in a direction intersecting (for example, a direction orthogonal to) the edge of the recognition target object to be detected. The subsequent intra-area projection luminance value calculation unit 103 calculates the luminance value L in the processing area 201 set by the processing area setting unit 102 as shown in FIG.
[0019]
In the subsequent edge candidate point storage unit 104, the luminance value L at each pixel position calculated by the intra-area projection luminance value calculation unit 103 and a predetermined value A are added, and the added value at each pixel position = L + A. A pixel position that substantially matches the luminance value L at each pixel position is calculated as an edge candidate pixel and stored in the memory. The subsequent edge point determination unit 105 integrates the number of edge candidate points calculated by the edge candidate point storage unit 104 for each pixel position, and determines a pixel having the largest number of edge candidate pixels as an edge pixel.
[0020]
FIG. 2A shows an image in the processing area 201 obtained through the image input unit 101 and the processing area setting unit 102. The processing area 201 includes a background area 203 and an electronic component area 204. FIG. 2B shows a luminance value L in the processing area 201 including the background area 203 and the electronic component area 204 calculated by the intra-area projection luminance calculation unit 103 in a two-dimensional graph of “pixel P versus luminance L”. Things.
[0021]
In FIG. 3, the edge candidate point storage unit 104 adds a predetermined value A to the luminance value L shown in FIG. 2, and compares the addition value at each pixel position = L + A with the luminance value L at each pixel position to substantially match. The pixel position is calculated as an edge candidate pixel, and then the number of edge candidate points calculated in the edge candidate point storage unit 104 is integrated for each pixel position by the edge point determination unit 105 (304 in the figure). FIG. 9 is a diagram illustrating a process of determining a pixel having the largest number of pixels as an edge pixel 305.
[0022]
Hereinafter, the operation will be described in detail. First, a plane image or a three-dimensional object whose one surface is not on the same plane as the background is imaged by a camera (not shown), and written and stored in one of a plurality of memories after A / D conversion. Read the most recent image written to memory to process the image. Next, an approximate center of the component image is obtained, and a processing area 201 including an electronic component area 204 and a background area 203 for detecting an edge from the center is set.
[0023]
In the processing area 201, the background area 203 and the electronic component area 204 are scanned to calculate the luminance value L (202) of each pixel. Conversely, by scanning the background area 203 from the electronic component area 204 and calculating the luminance value L of each pixel to determine an edge point, the center of the component can be determined.
[0024]
As shown in FIG. 3, from the calculated luminance value 202 of each pixel, a luminance value = L [0] + A obtained by adding the set value A to the luminance value L [0] of the 0th pixel P [0] is obtained. Compare with an arbitrary luminance value 202. Here, when the luminance value L is, for example, 8 bits (L = 0 to 255), it is assumed that A = about 40. If the addition value = L [0] + A is smaller than the maximum luminance value, a pixel position at which the addition value = L [0] + A is equal is set as an edge point candidate of the pixel P [0]. Similarly, edge point candidates for the next pixel P [1] are obtained, and edge point candidates for all pixels up to the final pixel P [max] are obtained. Then, the number of pixel positions that are considered as edge point candidates is integrated, and the pixel position with the largest number is determined as an edge point.
[0025]
Although the above embodiment has been described taking the case of recognizing components as an example, the present invention is also effective for detecting edges of what is printed or displayed on a two-dimensional plane, such as characters and figures.
[0026]
【The invention's effect】
As described above, according to the present invention, edges of components, characters, and the like can be obtained stably and can be accurately recognized.
[Brief description of the drawings]
FIG. 1 is a block diagram showing an embodiment of an edge detection device and a component recognition device according to the present invention; FIG. 2 is a graph showing luminance values near edges of components; FIG. 3 shows an edge detection process according to the present invention; FIG. 4 is an explanatory diagram showing a conventional edge detection method.
Reference Signs List 101 Image input unit 102 Processing area setting unit 103 Intra-area projection luminance value calculation unit 104 Edge candidate point storage unit 105 Edge point determination unit 201 Processing area 202 Luminance value 203 Background area 204 Electronic component area 304 Integrated value for each pixel position 305 Pixel with the largest number of edge candidate pixels (edge pixel)

Claims (6)

Adding means for adding a pixel value and a predetermined value at each pixel position of a plurality of pixels arranged in a direction intersecting the edge of the recognition target,
The addition value at each of the pixel positions added by the addition means is compared with the pixel value at each of the pixel positions of a plurality of pixels arranged in a direction intersecting with the edge of the object to be recognized, and a pixel that substantially matches is determined as an edge candidate. Edge candidate pixel calculating means for calculating as a pixel,
Edge pixel determining means for integrating the number of edge candidate pixels calculated by the edge candidate pixel calculating means for each pixel position, and determining a pixel having the largest number of edge candidate pixels as an edge pixel,
Edge detecting device having. Imaging means for imaging the component in a direction intersecting with the edge of the component,
A brightness value calculation unit that calculates a brightness value of each pixel position of a plurality of pixels arrayed in the intersecting direction captured by the imaging unit;
Adding means for adding a predetermined value and a luminance value at each pixel position calculated by the luminance value calculating means,
An edge candidate pixel for comparing an addition value at each of the pixel positions added by the addition means with a brightness value at each of the pixel positions calculated by the brightness value calculation means to calculate a pixel that substantially matches as an edge candidate pixel Calculating means;
Edge pixel determining means for integrating the number of the edge candidate pixels calculated by the edge candidate pixel calculating means for each of the pixel positions, and determining a pixel having the largest number of the edge candidate pixels as an edge of the component. ,
Edge detecting device having. 3. The edge detection device according to claim 2, further calculating an edge candidate pixel in a direction opposite to the direction intersecting with the edge, and calculating a direction intersecting with the edge and the number of edge candidate pixels in a direction opposite thereto with respect to each pixel position. A component recognition device that determines a pixel having the largest number of edge candidate pixels as an edge pixel, and determines a center surrounded by the edge pixel as a center of the component. An adding step of adding a pixel value and a predetermined value at each pixel position of a plurality of pixels arranged in a direction intersecting the edge of the recognition target,
The addition value at each of the pixel positions added in the adding step is compared with the pixel value at each of the pixel positions of a plurality of pixels arranged in a direction intersecting with the edge of the recognition target, and a pixel that substantially matches is determined as an edge candidate. An edge candidate pixel calculating step of calculating as a pixel;
Edge pixel determining step of integrating the number of the edge candidate pixels calculated in the edge candidate pixel calculating step for each pixel position, and determining a pixel having the largest number of the edge candidate pixels as an edge pixel,
Edge detection method. An imaging step of imaging the component in a direction intersecting with the edge of the component,
A brightness value calculation step of calculating a brightness value of each pixel position of a plurality of pixels arranged in the intersecting direction captured by the imaging step;
An adding step of adding a predetermined value and a luminance value at each of the pixel positions calculated by the luminance value calculating step;
An edge candidate pixel that calculates a pixel that substantially matches by comparing an addition value at each of the pixel positions added in the addition step with a brightness value at each of the pixel positions calculated in the brightness value calculation step as an edge candidate pixel A calculating step;
An edge pixel determining step of integrating the number of the edge candidate pixels calculated in the edge candidate pixel calculating step for each of the pixel positions, and determining a pixel having the largest number of the edge candidate pixels as an edge of the component. ,
Edge detection method. The edge detection method according to claim 5, further calculating an edge candidate pixel in a direction opposite to the direction intersecting with the edge, and calculating a direction intersecting with the edge and the number of edge candidate pixels in a direction opposite thereto in each pixel position. And a pixel having the largest number of edge candidate pixels is determined as an edge pixel, and a center surrounded by the edge pixel is determined as a center of the component.

Images