JP2007226334A - Color image processing method and color image processing device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To ensure accuracy in extracting an object irrespective of the brightness of the object on an image. <P>SOLUTION: In extracting a lead tip part for inspection in a substrate inspection device using color highlight illumination, each color data of R, G. B constituting a color image to be processed is weighted with a predetermined coefficient to form a grayscale image with the sum total I of each weighted color data as density. Positive values are set to the coefficients of R, G which are color data suitable for extracting the lead tip part to be inspected, out of the respective coefficients, and a negative value is set to the B. After specifying a threshold value to the formed grayscale image by a discrimination analysis method, binarization processing with the threshold value is carried out to extract the lead tip part. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention relates to a method and apparatus for extracting an object from a color image including a predetermined object.

  The applicant has filed a number of applications for a substrate inspection apparatus in which an illumination method called “color highlight illumination” is introduced (for example, Patent Document 1). In this substrate inspection apparatus, by imaging the substrate while irradiating each color light of red, green, and blue toward the substrate from different elevation angles, the highly specular portion has a color corresponding to the inclination angle. Generate the color image represented. Then, an inspection region is set at a preset position on the color image, and the image in the region is binarized to extract a region to be inspected. Further, the suitability of the inspection site is determined by measuring the position, area, and the like of the extracted inspection site and comparing the measured value with a predetermined determination reference value.

JP-A-9-145633

In the substrate inspection apparatus described in Patent Document 1, four types of parameters are set in order to extract the color of the part to be inspected, and two types of threshold values, an upper limit value and a lower limit value, are set for each of these parameters. is doing. One of the four types of parameters is lightness data, and is represented by the sum BRT of each color data of R, G, and B. The other three parameters are ratios ROP, GOP, and BOP (for example, ROP = R / BRT) of the R, G, and B color data with respect to the brightness data.
In general, the threshold values corresponding to these parameters are set by a user while looking at a color image of a model of a region to be inspected or a guidance screen, as disclosed in Patent Document 1.

  According to the above-described substrate inspection apparatus, a color image in which color data representing the inclination angle of the part having a high specular reflectivity is generated is generated for the part to be inspected with high specular reflectivity. It can be cut and extracted. However, when the reflectance of the site to be inspected is lowered, the intensity of the color data representing the tilt angle is weakened, and the brightness is reduced accordingly, so that erroneous extraction may occur.

For example, when the tip of an IC lead is inspected, the value of R generally increases because the tip is flat. Therefore, among the parameters described above, the threshold values of BRT and ROP are set higher. As a result, the lead can be extracted.
However, when the reflectivity of the lead is lowered due to palladium plating or the like, the strength of R is lowered, so the threshold values of BRT and ROP must be set low. As a result, a portion having a relatively high R intensity in the background portion may be erroneously extracted as noise, leading to a decrease in lead inspection accuracy.

  Note that the above problem is not limited to the reflectance of the region to be inspected, and may occur even when the intensity of reflected light incident on the camera cannot be ensured because, for example, the illumination conditions are not appropriate.

  Next, knowledge and experience are required to set the threshold value for the binarization process, but it is necessary to finely adjust the threshold value according to the reflectance of the part to be inspected and the illumination conditions. For this reason, it is not easy for an expert to set an optimal threshold value.

  In this regard, in Patent Document 2 below, variations in constituent colors of the object (which are considered to be synonymous with the color data in this specification) are automatically used using images of a plurality of recognition object samples. Extraction is performed, and an optimum color extraction range is obtained based on the variation range. However, in this patent document 2, constituent colors are extracted based on colors with high appearance frequency on an image. Therefore, when the reflectance of an object is low and the difference in constituent colors from the background is small There is a possibility that variations in constituent colors become inaccurate and the threshold value cannot be obtained accurately. Further, since it is necessary to perform statistical processing using images of a plurality of samples, the processing becomes large.

JP-A-2004-213200

The present invention has been made paying attention to the above-mentioned problems, and has as its first object to ensure the extraction accuracy of an object regardless of the brightness level of the object on the image. To do.
A second object of the present invention is to greatly reduce the burden on the user by automatically setting a threshold value suitable for the above-described extraction processing so that binarization can be performed. .

In the color image processing method according to the present invention, the following first and second steps are executed in order to extract the object from a color image including a predetermined object.
In the first step, among a plurality of color data constituting the color image, a positive coefficient is set for color data suitable for extraction of the object, and a negative coefficient is set for other color data. For each pixel included in the color image, an operation for obtaining the sum of the color data weighted by these coefficients is executed to generate a gray scale image in which each pixel is associated with the sum obtained by the operation.
In the second step, the object is extracted by binarizing the grayscale image generated in the first step using a predetermined threshold value.

In the above, “a plurality of color data constituting a color image” is, for example, three types of image data of R, G, and B, but is not limited thereto.
“Color data suitable for object extraction” is color data that is generally considered to be dominant over other color data in an object on a color image, but the color with the highest ratio in the object on the color image. One data may be selected, or two or more color data may be selected. However, it is not desirable to select all color data. Further, even if the color data appears in a certain size on the object, it is desirable to remove the data that predominates in the vicinity of the object from the “color data suitable for extraction of the object”.

According to the calculation in the first step, color data suitable for extraction of an object (hereinafter referred to as “extraction color data”) is strengthened by a positive value coefficient, and the ratio in the sum is increased. Color data other than color data is weakened by a negative coefficient, and the ratio in the sum is reduced. Therefore, in the gray scale image, it is possible to increase the density difference between the image area in which the extracted color data is more dominant than the other color data (corresponding to the object) and the other image areas.
In the above-mentioned “other image areas”, the proportion of the extracted color data is small, and the extracted color data is dominant in addition to the image areas where the color data other than the extracted color data is dominant. Image areas that are lower than those in the object are also included.

Therefore, even for an object whose extracted color data value does not show a significant size due to low reflectance or low illumination intensity, the density difference with respect to the background is large on the grayscale image. Appears as an image area. In addition, since the negative value coefficient can reduce the influence of color data other than the extracted color data, even if the background contains a portion where the color data other than the extracted color data is large, It is possible to prevent the concentration of the portion from increasing.
Therefore, regardless of the brightness of the object on the color image, the object can be accurately separated from the background portion.

  In a preferred aspect of the above method, the coefficient corresponding to each color data is set so that the result of the calculation for the pixels constituting the background of the object is a value that approximates zero. According to this setting, on the gray scale image, the density value other than the object is almost 0, so that the object and the background can be accurately separated.

  In another preferred aspect, in the second step, a threshold for binarization is specified based on a density distribution state of the gray scale image, and the gray scale image is determined based on the specified threshold. Is binarized.

  In the above aspect, for example, a density histogram of the gray scale image is created based on a known method such as a discriminant analysis method, and a threshold value for binarization is specified based on the distribution state of the histogram. In this case, in the first step, a grayscale image having a large density difference between the object and the background is generated, so that a threshold suitable for extraction of the object can be easily specified.

  In the color image processing method according to the present invention, the entire color image generated by imaging the object may be processed, or a predetermined area (position of the object) in the entire image may be processed. Only the color image in the area) set based on the information may be processed.

  Next, a color image processing apparatus according to the present invention includes an image input means for inputting a color image obtained by imaging a predetermined object, and the object among a plurality of color data constituting the color image. For each pixel included in a predetermined area in the color image input by the image input means by setting a positive coefficient for the color data suitable for the extraction and a negative coefficient for the other color data. Generating a gray scale image in which the calculation means for calculating the sum of the color data weighted by each coefficient is associated with each pixel in the predetermined area and the sum obtained by the calculation of the calculation means And a threshold value setting means for setting a threshold value for binarizing the gray scale image based on the density distribution state of the gray scale image generated by the image generation means And a binarizing means for binarizing the gray-scale image by the threshold set by the threshold value setting means comprises.

  According to the color image processing apparatus having the above-described configuration, it is possible to accurately extract an object by executing the above-described color image processing method. Therefore, it is possible to ensure the accuracy of the measurement and inspection performed thereafter. In addition, every time a color image is input, a threshold suitable for a grayscale image generated from the color image is automatically set, so that it is not necessary to set a threshold in advance.

  The area to be processed by the image generation means may be a partial area in the input color image, but is not limited to this, and may be the entire input color image.

According to the present invention, instead of processing a color image of an object, the effect of the intensity of color data suitable for the extraction of the object is increased, and the influence of the intensity of other color data is reduced. Since the scale image is processed, even if the object on the color image is not sufficiently bright due to the reason that the reflectance of the object itself is small or the illumination is insufficient, the object Objects can be extracted with high accuracy.
Also, by performing extraction processing using the gray scale image as described above, it is possible to perform binarization after automatically setting a threshold suitable for extraction, greatly reducing the burden on the user. can do.

FIG. 1 shows a configuration example of a substrate inspection apparatus to which the present invention is applied.
The board inspection apparatus according to this embodiment performs an inspection at each component mounting position on a substrate that has undergone a component mounting process in a component mounting board manufacturing line, and includes a controller 1, a camera 2, a substrate stage 3, The illumination device 4, the input unit 5, the monitor 6, and the like are included.

  The camera 2 generates a color still image, and is fixedly arranged above the substrate stage 3 with the imaging surface facing downward. On the substrate stage 3, a table 3a (shown in FIG. 2) for horizontally supporting the substrate to be inspected, and the table 3a are moved in the X direction (substrate length direction) and Y direction (substrate width direction). A moving mechanism (not shown) is included. The illumination device 4 is for illuminating a substrate to be inspected.

The controller 1 is provided with an image input unit 12, an XY stage control unit 13, an illumination control unit 14, a memory 15, an inspection result output unit 16 and the like in addition to a computer control unit 11.
The image input unit 12 includes an interface circuit for the camera 2 and an A / D conversion circuit. The XY stage control unit 13 controls the movement of the substrate stage 3, and the illumination control unit 14 adjusts the turning on / off operation of the lighting device 4.

The memory 15 stores a program in which a series of processing procedures for inspection are described, an inspection data file, and the like.
The inspection data file stores conditions for setting the inspection region (region position and size) to be set on the substrate, determination reference values for determining the quality of the region to be inspected, and the like. Further, in this embodiment, since the substrate is divided into a plurality of imaging target areas for imaging, data indicating the set position of each imaging target area is registered in the inspection data file. A ratio of coefficients for generating a gray scale image to be described later is also stored in the inspection data file.

  The control unit 11 controls the movement of the substrate stage 3 via the XY stage control unit 13, thereby aligning the camera 2 with each imaging target region in order and taking an image. The color image generated here is input to the control unit 11 via the image input unit 12 and stored in its internal memory (RAM). The control unit 11 sets an inspection region for each inspected portion of the color image stored in the RAM based on the setting conditions in the inspection data file, and in that region, extracts, measures, Each process of determination is performed sequentially.

  When the determination processing for all components on the board is completed, the control unit 11 collects these determination results and determines whether the board is good or bad. Information regarding the final determination result or a defective part when it is determined to be defective is given to the inspection result output unit 16 and then output from the inspection result output unit 16 to an external device (not shown).

  In the above-described board inspection apparatus, when inspecting lead parts such as IC and QFP, an inspection area is set for each lead to inspect whether the leading end of the lead is bent or lifted. .

FIG. 2 shows a state in which the lead component 8 on the substrate 7 is imaged by the camera 2 and the illumination device 4 together with the arrangement relationship of the light sources of the camera 2 and the illumination device 4. In the figure, reference numeral 80 denotes a lead, which has a root portion 82 attached to the component main body, a tip portion 81 soldered to the land of the substrate 7, and an intermediate portion 83 connecting these. Reference numeral 9 denotes solder for connecting the tip 81 and the land.
The illumination device 4 includes three annular light sources 4R, 4G, and 4B (hereinafter referred to as “red light source 4R”, “green light source 4G”, and “blue light source 4B”) that emit red, green, and blue color lights, respectively. .) Is deployed with each center aligned. The camera 2 is disposed above the illumination device 4 so that the optical axis is aligned with the center of each light source 4R, 4G, 4B.

  The red light source 4 </ b> R has the smallest diameter among the three light sources and is disposed at a position closest to the camera 2. The blue light source 4 </ b> B has the largest diameter among the three light sources and is disposed at a position closest to the substrate 7. The green light source 4G is disposed between the light source 4R and the light source 4B, and the diameter thereof is set to be between the diameter of the light source 4R and the diameter of the light source 4B.

With this configuration, red, green, and blue color lights are irradiated onto the substrate 7 from different elevation angles. At portions having high specular reflectivity on the substrate (soldering portion 9, lead 80, etc.), these color lights are regularly reflected, but the type of reflected light incident on the camera 2 is the inclination angle of the position where the light is irradiated. It will be different depending on.
For example, in the case of the lead 80, the top end portion 81 and the root portion 82 have a substantially flat upper surface, so that reflected light for red light is mainly incident on the camera 2. Further, since the inclination is steep at the intermediate portion 83 and the edge 81a of the tip portion 81, reflected light for blue light is mainly incident. Further, at the portion corresponding to the boundary between the flat surface and the steep slope, the ratio of incident reflected light to green light is increased.

FIG. 3 schematically shows a color image 100 obtained by imaging a part of the lead component 8 generated under illumination by the illumination device 4. For convenience, the same reference numerals as those in FIG. Each part 81, 82, 83 of the lead 80 is represented by a pattern (corresponding to the one attached to the light sources 4R, 4G, 4B in FIG. 2) corresponding to the color data dominant on the image. However, since the area where the green color appears remarkably is very small, the illustration is omitted here.
Reference numeral 91 denotes an image of a land not shown in FIG. In addition to the region where R is dominant, the region where G and B are dominant depending on the tilted state also appears in the soldering portion 9, but since it is not an extraction target of this embodiment, it does not correspond to R, G, B. This is represented by the pattern.

  In this embodiment, in order to inspect the state of the tip portion 81 of the lead 80 (hereinafter referred to as “lead tip portion 81”), an inspection region R is set for each lead tip portion 81, and each inspection region R is set. The color image is converted into a gray scale image and binarized. In the binarization process, a threshold value is automatically extracted using a discriminant analysis method.

Here, a method of generating the gray scale image will be described in detail.
In this embodiment, for each pixel included in the inspection region R, color data R, G, B representing the color of the pixel (hereinafter, each color data is referred to as “red data R”, “green data G”). , “Blue data B”) is applied to the following equation (1) to generate a grayscale image having the calculated I value as the density value.
_{I = C R × R + C} G × G + C B × B ··· (1)

Among the coefficients C _R , C _G , and C _{B in} the above equation (1), positive values are set for the coefficients of the color data representing the desired inclination state of the lead tip 81, and negative for the other color data coefficients. The value of is set.
As shown in FIG. 3, the red data R is dominant at the lead tip portion 81, but the ratio of the green data G is also increased at a position close to the boundary with the intermediate portion 83 and the edge 81a. In consideration of this point, in this embodiment, a coefficient value C _R for the red data R, the coefficient value C _G corresponding to the green data G, and then to set a positive value.

Further, in this embodiment, the ratio of the positive coefficients C _R and C _G is matched with the ratio of R and G appearing in the image of the lead tip portion 81 in the desired inclined state. This ratio is extracted in advance from the color image of the lead tip 81 having a sufficiently high reflectance and stored in the inspection data file.

Further, in this embodiment, when the expression (1) is applied to a color image (hereinafter referred to as “background image”) of the ground surface of the substrate 7 (the surface of the substrate 7 on which the land 91 or the like is not formed), The values of the coefficients C _R , C _G , and C _B are determined so that the value of I becomes zero.

Hereinafter, the ratio of the coefficients C _R and C _G registered in the inspection data file is 3: 1, and the image data of the background image is R = 45, G = 60, and B = 45 (all of which have an 8-bit configuration). It will be explained under the assumption that it is data.

First, assuming that C _G = a (a> 0), the calculation result of the expression (1) based on the image data of the background image is set to 0.
3a × 45 + a × 60 + C _B × 45 = 0 (2)

If the above formula (2) is arranged, C _B = −4.33a.
Here, if a = 3 so that C _B becomes a value close to an integer, C _R = 9 and C _G = 3. Further, when the decimal point of C _B is raised, C _B = −13.

According to the coefficients C _R , C _G , and C _B described above, the red data R and the green data G that indicate the inclination state of the surface of the lead tip 81 are strengthened, while the blue data B that does not indicate the inclination state is negative. It is weakened by the coefficient C _B. Thus, the lead tip 81 of the color image, the absolute value of equation (1) while the positive value in the right-hand side _{(C R × R + C G} × G) becomes large, negative value (C _B × B) is Since it becomes small, the density | concentration I is considered to become a high value.

In addition, in the lead 80 having low reflectance such as the lead 80 plated with palladium, the values of the red data R and the green data G of the lead tip 81 in the color image are low, but according to the above equation (1), since the value of R and G is emphasized by the coefficient C _R, C _G, it is possible to secure a certain value to the concentration I.

On the other hand, regarding the ground surface portion of the substrate, the density I is considered to show a value close to 0 unless there is a large variation in color. In the blue predominant steep such as the intermediate section 83, the value of R and G originally small, positive coefficient C _R, the value of the weighted not be so large due to C _G. On the other hand, B is increased in the minus direction by weighting with the negative coefficient C _B , so that the calculation result of the equation (1) is considered to be a value close to 0 or a negative value. When I becomes a negative value, it is considered that I = 0. Therefore, according to the inspection region R shown in FIG. 3, the color image of the portion other than the lead tip 81 is converted to 0 or a value close to 0 unless noise with a high R or G value is generated. It is thought that it is done.

  As described above, according to the gray scale image according to the equation (1), even when the reflectance of the lead tip portion 81 to be extracted is low, it is significant between the image of the lead tip portion 81 and the other portions. It becomes possible to set the density difference. Therefore, since a density histogram with high bimodality can be created from this gray scale image, a threshold value for binarization processing is automatically extracted using a discriminant analysis method, and binarization processing with high accuracy is performed. It becomes possible to do.

FIG. 4 is a graph showing the density distribution of each color data R, G, and B in the color image and the density distribution on the gray scale image for the lead 80 having a high reflectance and the lead 80 having a low reflectance. It is. Each graph shows the density distribution along the length direction of the lead tip 81, and the vertical axis shows density (up to 255 gradations), and the horizontal axis shows pixel positions.
Note that the coefficients C _R , C _G , and C _B for generating the gray scale image are C _R = 9, C _G = 3, and C _B = −13 based on the above-described specific example, and the result of the calculation Is normalized.

In the graph 1, the red data R becomes much more dominant than the other color data G and B in a wide central region, and peaks of G and B appear in the vicinity thereof. In addition, at both end positions in the R dominant region, G shows a larger change than B. It can be said that these phenomena accurately reflect the inclined state of the lead tip portion 81.
Reflecting these values of R, G, and B, the value of I increases in the range where the values of R and G are high, and outside thereof, it is substantially zero except for the portion indicated by the arrow f1.

Even in the graph 2, R is dominant in the central portion, and G and B peaks appear in the vicinity thereof. However, compared to the graph 1, the peak of any color data is low and the degree of variation is small. .
However, two large peaks appear in the concentration distribution of I. In the region sandwiched between these peaks, a value of at least about 60 is secured, while in other portions, it is almost 0 except for the portion indicated by the arrow f2.

  The portions indicated by arrows f1 and f2 in the graphs 1 and 2 are considered to be noise generated because the value of R is slightly increased. However, even in the graph 2 in which the value of I is not so high, a considerable difference is recognized between the noise portion and the range corresponding to the lead tip portion 81. Therefore, in the density histogram for discriminant analysis, the noise It is unlikely that the value of I at and the value of I at the lead tip 8 are included in the same peak.

Next, the effect of changing the value of the coefficient C _R for the red data R, consider.
Graphs 3 and 4 of FIG. 5, when processing the same color image as in FIG. 4, by changing the coefficients _{C R} for the red data R from 9 8 (other coefficients _C G, _{C B} is the same.), The example which calculated the density | concentration I of the gray scale image is shown. Also in this example, for a lead having a high reflectance, the value of I is increased in a large range of R and B corresponding to the lead tip 81 (see graph 3).

However, in the graph 4 corresponding to the lead having a low reflectivity, there is a portion where the value of I drops to 0 in the central portion that should correspond to the lead tip portion 81. This, R, since the difference between the B small, by having a small positive coefficient C _R, the influence of the subtraction amount by negative coefficient C _B is considered to be because the increased.

Graphs 5 and 6 of FIG. 6, when processing the same color image as in FIG. 4, the coefficients _{C R} for the red data R changed from 9 to 10 (other coefficients _C G, _{C B} is the same.), The example which calculated the density | concentration I of the gray scale image is shown. In this example, in both the graphs 5 and 6, the value of I in the portion corresponding to the lead tip portion 81 is high, but the noise values indicated by the arrows f1 and f2 are also larger than those in the graphs 1 and 2. .

According to FIG. 5 and 6, when adjusting the coefficients C _R, adjusted to lower values is considered undesirable. Also, when raising the value, it is necessary to take into consideration that noise in the background portion is emphasized.

In the above description, in order to determine the ratio of the positive coefficients C _R and C _G , the R and G color data are extracted from the color image of the lead tip 81 having a high reflectance. It is not limited to. For example, the area of the region with the strongest R and the area of the region with the strongest G are measured from the image of the lead tip 81 having a high reflectivity, and the ratios of the coefficients C _R and C _G are determined based on the area ratio. May be. If the user knows the ratio between the R dominant region and the G dominant region empirically, the ratio of the coefficients C _R and C _G may be determined based on the knowledge.
Alternatively, the length, width, inclination angle, and the like of the lead tip 81 are measured and incident on the camera 2 from the lead tip 81 based on the measured values and the angle range of illumination light from the light sources 4R and 4G. The ratio of the reflected light of R and G may be calculated, and the ratio of the coefficients C _R and C _B may be determined according to the ratio.

FIG. 7 shows a flow of processing executed by the substrate inspection apparatus. Here, in order to simplify the description, the flow limited to the inspection of the lead tip portion 81 is shown.
First, in the first ST 1 (ST is an abbreviation of “step”. The same applies to the following), the substrate 7 to be inspected is carried into the substrate stage 3. In ST2, the field of view of the camera 2 is aligned with an area where the ground surface of the substrate 7 can be imaged, and imaging is performed. Thereby, the above-described background image is generated.

In ST3, R, G, and B values are extracted from the background image. This extraction process may be performed on one pixel, but the values of a plurality of pixels may be extracted for R, G, and B, and the average value thereof may be obtained.
In the next ST4, the coefficient _C R from the inspection data file, reads the ratio of _{C G,} R of the background image extracted by the this ratio ST3, G, coefficients using the values of B _{C R, C} G, Determine the value of C _B. This determination method is as described above with reference to the specific example using the expression (2).

  Thereafter, in ST5, the position of the substrate stage 3 is adjusted, the field of view of the camera 2 is matched with the lead 80 to be inspected, and imaging is performed. As a result, a color image as shown in FIG. 2 is generated.

In ST6, an inspection area is set on the color image generated in ST5.
In ST7, Expression (1) is executed for each pixel in the inspection region using each set coefficient to obtain the density I. Further, a gray scale image of the inspection area is generated by associating the calculated density I with the coordinates of the pixel to be processed.

In ST8, a density histogram is created from the gray scale image, and a density I corresponding to a predetermined position between two peaks appearing in the histogram is set as a threshold value.
In the next ST9, the gray scale image is binarized by the threshold set in ST8. Thereby, the lead tip portion 81 is extracted.

  In ST10, the extracted lead tip 81 is subjected to processing for measuring the position and length of the tip edge. Further, the quality of the lead tip 81 is determined by comparing the measurement result with a predetermined determination reference value. This determination result is temporarily stored in the RAM.

Hereinafter, each step of ST7-10 is performed with respect to all the test | inspection area | regions R contained in an image. Although not shown here, when one part is imaged in multiple times, or when there are a plurality of parts to be inspected, the camera 2 is matched to each imaging target area in order, and each imaging target area is The processing of ST5 to 11 is executed.
In this way, when the processing for all the inspection regions R on the substrate 7 is completed, ST11 becomes “YES”, and after the inspection result is output in ST12, the inspection is ended.

  In the example of FIG. 7, the background image generation process (ST2) and the coefficient determination process (ST3) are executed for each substrate 7 to be inspected. However, the present invention is not limited to this, and these steps are executed. May be applied only to the first substrate 7 to be inspected, and the coefficient value obtained for this substrate 7 may be applied to the subsequent substrates 7.

It is a block diagram which shows the structure of the board | substrate inspection apparatus to which this invention was applied. It is explanatory drawing which shows the imaging state at the time of the test | inspection of a lead component with the arrangement | positioning relationship between a camera and an illuminating device. It is the figure which showed the color image of lead components typically. It is a graph which shows the distribution state along the length direction of a lead front-end | tip part about the density | concentration I of R, G, B in a color image, and a gray scale image. 5 is a graph showing a result of calculating a grayscale image density value I by reducing the coefficient value for R by one for a color image similar to FIG. 4. FIG. 5 is a graph showing a result of calculating a density value I of a grayscale image by increasing the coefficient value for R by one for a color image similar to FIG. 4. It is a flowchart which shows the flow of the process at the time of a test | inspection.

Explanation of symbols

1 Controller 2 camera 4 lighting device 11 control unit 15 memory 81 lead tip portions 100 color images R, G, B color data _{C R, C G,} the density value of _{C B} coefficients I grayscale image

Claims (4)

A method of extracting the object from a color image including a predetermined object,
Of the plurality of color data constituting the color image, a positive value coefficient is set for color data suitable for extraction of the object, and a negative value coefficient is set for the other color data. A first step of generating a grayscale image in which each pixel is associated with the total obtained by the calculation, by performing an operation for obtaining a sum of color data weighted by these coefficients for each pixel included in
A second step of extracting the object by binarizing the grayscale image generated in the first step with a predetermined threshold;
A color image processing method comprising: executing the color image processing method.   2. The color image processing method according to claim 1, wherein the coefficient for each color data is set such that a result of the calculation for pixels constituting a background of the object becomes a value close to 0. 3.   In the second step, a threshold value for binarization is specified based on a density distribution state of the gray scale image, and the gray scale image is binarized by the specified threshold value. Item 4. A method for processing a color image according to Item 1. Image input means for inputting a color image obtained by imaging a predetermined object;
Of the plurality of color data constituting the color image, a positive value coefficient is set for color data suitable for extraction of the object, and a negative value coefficient is set for the other color data. Calculating means for executing a calculation for obtaining a sum of color data weighted by each coefficient for each pixel included in a predetermined area in the color image input by the means;
Image generation means for generating a grayscale image in which each pixel in the predetermined region is associated with the sum obtained by the calculation of the calculation means;
Threshold setting means for setting a threshold for binarizing the gray scale image based on the density distribution state of the gray scale image generated by the image generating means;
A color image processing apparatus comprising: binarizing means for binarizing the gray scale image with a threshold set by the threshold setting means.

Images