CN105044114B - A kind of electrolytic capacitor appearance packaging defect image detecting system and method - Google Patents

Abstract

The invention discloses an image detection system for appearance packaging defects of electrolytic capacitors, which includes a feed tank unit, an upper surface image acquisition unit, a lower surface image acquisition unit, a side surface image acquisition unit, a central processing unit, a display unit and rejecting unqualified products unit; the feed trough unit is fed by the vibrating plate to the capacitor at intervals according to the teeth, the outlet of the vibrating plate is the feeding trough, the vibrating plate is a concave plate with a spiral line, and the feeding trough is defined by two rectangular plates Transmission transparent slot, there is an electrolytic capacitor detection station in the middle of the transmission transparent slot, and the pins of the electrolytic capacitor are inserted at the lower end of the transparent slot; the transparent slot has three stations, which are obtained by image acquisition on the upper surface, image acquisition on the side surface and image acquisition on the lower surface Image processing is performed on the three-direction images of electrolytic capacitors to identify unqualified products with defective appearance.

Description

An image detection system and method for appearance packaging defects of electrolytic capacitors

technical field

The invention relates to a defect detection method for the exterior packaging of electrolytic capacitors, in particular to a defect detection system and method for the exterior packaging of electrolytic capacitors based on digital image processing.

Background technique

At present, the defect inspection of the external packaging of electrolytic capacitors still relies on manual visual judgment, that is, to observe whether the round edges of the plastic packaging on the upper and lower surfaces of the capacitor are regular, whether the wrapping is aligned with the edge of the capacitor, and whether there are protruding edges Damage, whether the position of the positive and negative symbols on the surface of the capacitor is correct, whether the positive and negative symbols correspond to the long and short pins of the capacitor, and whether the model information is printed correctly, etc. Manual one-by-one detection has the problems of large labor demand, slow speed and low efficiency, which seriously affects the productivity and economic benefits of enterprises.

Contents of the invention

In view of the above technical problems, the object of the present invention is to provide an automatic detection system and method for capacitive appearance packaging defects with simple structure and capable of replacing manual operation, so as to improve detection efficiency.

The present invention adopts the following technical solutions to solve the above-mentioned technical problems: an image detection system for defects in the appearance and packaging of electrolytic capacitors, which is characterized in that it includes a feed tank unit, a light source unit, an upper surface image acquisition unit, a photoelectric switch unit, and a light source unit. The lower surface image acquisition unit, the three light source units, the three photoelectric switch units, the side surface image acquisition unit, the central processing unit, the display unit and the rejecting unit for unqualified products; Feeding at intervals, the outlet of the vibrating plate is the feeding trough 1-2, the vibrating plate is a concave plate with a spiral line, the feeding trough is a transmission transparent trough defined by two rectangular plates, and an electrolytic capacitor 5 is installed in the middle of the transmission transparent trough for detection In the station, the 6 pins of the electrolytic capacitor are inserted into the lower end of the transparent slot; there are three stations on the transparent slot, and the light source unit 1 is located directly above the image acquisition unit 2 on the upper surface of the station one, providing light for the image acquisition unit on the upper surface, The upper surface image acquisition unit collects the appearance image of the upper surface of the capacitor after the first photoelectric switch unit detects the capacitor; the second light source unit is located directly below the lower surface image acquisition unit 3 at station 2, providing the lower surface image acquisition unit Light, the lower surface image acquisition unit collects the appearance image of the lower surface of the capacitor after the second photoelectric switch unit detects the electrolytic capacitor; the third light source unit is located on the side of the side surface image acquisition unit of station three, and is the side surface image acquisition unit Provide light, and the side surface image acquisition unit continuously collects the appearance image of the side surface of the capacitor after the third unit of the photoelectric switch detects the capacitor. The third station is a rotating station at the same time, and the rotating stations 1-3 drive the pins to make the capacitor rotate for a circle. , the photoelectric switch three unit detects that the capacitor enters the station three and triggers the rotation, that is, the stepper motor rotating plate on the station on the feeding chute drives the capacitor to rotate for one circle. The input terminal of the central processing unit receives image signals to analyze and process the collected three-sided images. The display unit connected to the central processing unit displays the collected images and processing results. Products are rejected.

Further, color image grayscale, image denoising, image enhancement, region segmentation, image normalization, and defect detection are performed on the upper and lower surface images respectively; color image grayscale, image denoising, image enhancement, and image stitching are performed on the side image , region segmentation, image normalization, and defect detection; the grayscale of the color image is used to convert the color image acquired by the camera into a grayscale image, and the image denoising is used to remove noise from the grayscale image, The image enhancement is used to enhance the contrast of the image, the region segmentation is used to segment the region to be detected, and the image normalization is used to perform normalization processing such as rotation, cropping and scaling of the image to be detected , the defect detection is used to identify the defect between the normalized image to be detected and the existing standard normalized image template, and the image mosaic is used to mosaic the capacitive acquisition image that rotates one revolution into a complete side image ;

Specific steps are as follows:

Step 001. As soon as the photoelectric switch detects the capacitance, it triggers the camera on the upper surface image acquisition unit to capture the image;

Step 002. Convert the upper surface color image into a grayscale image;

Step 003. Filter the upper surface grayscale image to remove image noise;

Step 004. Perform image enhancement on the denoised upper surface grayscale image to enhance the contrast of the image;

Step 005. Perform region segmentation on the enhanced upper surface image, specifically including the following steps:

Step 0051. Utilize the sobel edge detection operator to perform edge detection on the enhanced upper surface image;

Step 0052. Utilize the hough transform to detect the appearance edge circle curve of the upper surface of the capacitor on the edge of the upper surface image;

Step 0053. Mark the area where the circular curve on the appearance edge of the connected capacitor is located, and obtain the segmented upper surface area to be detected;

Step 006. In particular, if there is character mark information on the package on the upper surface, character mark detection is required, including the following steps:

Step 0061. Perform horizontal or vertical grayscale projection on the segmented upper surface area to be detected, and segment the area where the specification information of the upper surface appearance of the capacitor is located;

Step 0062. Rotate the upper surface to-be-detected area obtained in step 0053 to the vertical direction by using the angle of the line where the upper-left corner and the lower-left corner pixel in the image of the upper-surface specification information to-be-detected area deviates from the vertical direction; wherein the upper-left corner pixel is from From left to right, the first pixel found from top to bottom, the bottom left pixel is the first pixel found from left to right, from bottom to top.

Step 007. Normalize the segmented region image, specifically including the following steps:

Step 0071. Cut the image of the area to be detected on the upper surface rotated to the vertical direction into a square image with 5 pixels in the upper, lower, left, and right sides of the circle; it is to rotate the image in any direction that has been captured by software to make it consistent with the orientation of the standard template image , see step 0062 for the specific rotation angle. The cropping should also be consistent with the standard template. In order not to lose any information, small-scale expansion is done.

Step 0072. Scale the cropped image of the area to be detected on the upper surface to a specified calibration image of a specified pixel size (such as 160×160);

Step 008. Carry out an XOR operation on the normalized upper surface to-be-detected area image and the standard upper surface template image in the image database (the acquisition process is also through steps 001-007). If the calculation result exceeds the threshold, it is determined that the upper surface is not qualified.

Step 009. The second photoelectric switch detects the capacitance, and triggers the camera of the image acquisition unit on the lower surface to capture the image;

Step 0010. Convert the lower surface color image into a grayscale image;

Step 0011. Filtering the lower surface grayscale image to remove image noise;

Step 0012. Perform image enhancement on the denoised lower surface grayscale image to enhance the contrast of the image;

Step 0013. Perform region segmentation on the enhanced lower surface image, specifically including the following steps:

Step 00131. Utilize the sobel edge detection operator to perform edge detection on the enhanced lower surface image;

Step 00132. Use hough transform to detect the edge circle curve of the appearance edge of the capacitor lower surface on the image edge of the lower surface; Step 00133. Mark the area where the circle curve of the appearance edge of the connected capacitor is located, and obtain the segmented lower surface area to be detected. The mark here is the software in the field of image processing Use a rectangular frame to frame the meaning of the circle boundary;

Step 0014. In particular, if there is character identification information on the lower surface packaging, character identification detection (also can be used as the center or edge mark of the area) is required, which is characterized in that it includes the following steps:

Step 00141. Perform horizontal or vertical grayscale projection on the segmented lower surface area to be detected, and segment the area where the specification information of the lower surface appearance of the capacitor is located;

Step 00142. Rotate the area to be detected on the lower surface obtained in step 00133 to the vertical direction by using the angle of the line where the upper left corner and the lower left corner pixel in the image of the area to be detected in the specification information of the lower surface deviates from the vertical direction; wherein the upper left corner pixel is from From left to right, the first pixel found from top to bottom, the bottom left pixel is the first pixel found from left to right, from bottom to top.

Step 0015. Normalize the segmented region image, specifically including the following steps:

Step 00151. Cutting the image of the area to be detected on the lower surface rotated to the vertical direction into a square image with 5 pixels in the top, bottom, left, and right sides of the circle;

Step 00152. Scale the cropped image of the area to be detected on the lower surface to a specified calibration image of a specified pixel size (such as 160×160);

Step 0016. Carry out XOR operation with the normalized image of the area to be detected on the lower surface and the standard lower surface template image in the image database (the acquisition process is also through steps 009-0015), and the judgment that the result of the operation exceeds the threshold is that the lower surface is not qualified.

Step 0017. The photoelectric switch 3 detects the capacitor, triggers the side surface image acquisition unit camera to capture the image, and simultaneously triggers the rotating plate on the feeding trough to drive the capacitor to rotate for one revolution;

Step 0018. Convert a plurality of side surface color images into grayscale images;

Step 0019. Filtering and removing image noise on multiple side surface grayscale images respectively;

Step 0020. Perform image enhancement on the multiple side surface grayscale images after denoising to enhance the contrast of the image;

Step 0021. Perform image mosaic on the enhanced side surface images, specifically including the following steps:

Step 00211. Extract feature points from the left and right ends of the enhanced side images;

Step 00212. Matching the feature points extracted from the left and right ends;

Step 00213. Perform image fusion according to the matching points to obtain the entire side image after splicing.

Step 0022. Perform horizontal or vertical grayscale projection on the entire side image obtained by splicing, and segment the area where the specification information of the surface appearance of the capacitor side is located;

Step 0023. Rotate the area to be detected on the side surface to the vertical direction by using the angle of the straight line where the pixels in the upper left corner and the lower left corner of the image of the region to be detected in the side surface specification information deviates from the vertical direction; wherein the pixels in the upper left corner are from left to right, The first pixel found from top to bottom, the bottom left pixel is the first pixel found from left to right and bottom to top.

Step 0024. Normalize the segmented region image, specifically including the following steps:

Step 00241. Crop the image of the area to be detected on the side surface rotated to the vertical direction into a square image of 5 pixels from the edge, up, down, left, and right;

Step 00242. Scale the cropped image of the area to be detected on the side surface to a specified calibration image with a specified size of pixels (such as 160×160);

Step 0025. Carry out XOR operation with the normalized side surface to-be-detected area image and the standard side surface template image in the image database (its acquisition process is also through steps 0017-0024), and the judgment that the operation result exceeds the threshold is that the lower surface is not qualified.

The standard upper surface template image is produced by qualified products using steps 001-007, the standard lower surface template image is produced by qualified products using steps 009-0015, and the standard side template image is produced by qualified products using steps 001-001. 0017~0024 are processed and generated.

The direction of the capacitor long pin detected by the photoelectric switch can be combined to determine the direction of the positive symbol in the appearance image.

The display unit can display the whole visual inspection process.

Unqualified products can be added according to the needs of the defect category display unit display function and the removal of different defect categories.

The number of images for image stitching is selected according to the field of view of the side surface acquisition unit camera. The image enhancement is used to enhance the contrast of the image, the region segmentation is used to segment the region to be detected, and the image normalization is used to perform normalization processing such as rotation, cropping and scaling of the image to be detected , the defect detection is used to identify the defect between the normalized image to be detected and the existing standard normalized image template, and the image mosaic is used to mosaic the capacitive acquisition image that rotates one revolution into a complete side image .

The beneficial effect of the present invention is that the present invention proposes an image detection system and method for defects in the exterior packaging of electrolytic capacitors. Compared with the existing manual operation technology, the above technical scheme has the following beneficial technical effects: it is first applied in the appearance detection of capacitors , It can realize automatic and intelligent detection of whether there are defects or errors in the upper and lower surface packaging edges and side model information of capacitor product packaging. The detection processing algorithm is effective, fast, and high detection rate.

(1) An image detection system for electrolytic capacitor appearance packaging defects designed by the present invention has simple structure, low cost, easy implementation, and complete functions, which can realize automatic detection of capacitor appearance packaging defects and effectively improve detection efficiency.

(2) A kind of electrolytic capacitor exterior packaging defect image detection method designed by the present invention is specially designed for the characteristics of capacitance appearance detection. The algorithm is practical and effective. Under the condition of not increasing the complexity of the algorithm, the recognition rate is high and the operating speed Fast and highly reliable.

Description of drawings

Fig. 1 is the functional block diagram of the image detection system of the appearance packaging defect of electrolytic capacitor of the present invention;

Fig. 2 is a functional block diagram of an image detection method for an appearance and packaging defect of an electrolytic capacitor according to the present invention;

Figure 3 is a working diagram.

Detailed ways

The specific implementation manners of the present invention will be further described in detail below in conjunction with the accompanying drawings.

As shown in Figures 1 and 3, the present invention designs an image detection system for appearance packaging defects of electrolytic capacitors, including a feed tank unit 1, a light source unit 1, an upper surface image acquisition unit 3, a photoelectric switch unit 1, a light source unit 2, Lower surface image acquisition unit 4, three light source units, three photoelectric switch units, side surface image acquisition unit 5, central processing unit, display unit and unqualified product rejection unit. The feed trough unit uses the vibrating plate 1-1 to feed the capacitors at intervals according to the teeth. The outlet of the vibrating plate is the feeding trough 1-2. The vibrating plate is a concave plate with a spiral line, which can automatically discharge the electrolytic capacitors. The pins of the electrolytic capacitors are arranged downwards. The feeding trough is a transparent trough defined by two rectangular plates. The electrolytic capacitor 5 is placed in the middle of the transparent trough, and the 6 pins of the electrolytic capacitor are inserted into the lower end of the transparent trough; there are three stations on the transparent trough. , the light source unit 1 is located directly above the upper surface image acquisition unit 2 of station 1, and provides light for the upper surface image acquisition unit, and the upper surface image acquisition unit performs an appearance image on the upper surface of the capacitor after the photoelectric switch unit 1 detects the capacitor Acquisition; the second light source unit is located directly below the lower surface image acquisition unit 3 of station two, and provides light for the lower surface image acquisition unit. The lower surface image acquisition unit detects the electrolytic capacitor on the lower surface of the capacitor after the second photoelectric switch unit detects Perform appearance image acquisition; the third light source unit is located on the side of the side surface image acquisition unit of station three, providing light for the side surface image acquisition unit, and the side surface image acquisition unit detects the capacitor after the third photoelectric switch unit detects the capacitor. The appearance image is continuously collected, and the third station is a rotating station at the same time. The rotating station 1-3 drives the pins to make the capacitor rotate for a circle. The stepper motor rotating plate on the bit drives the capacitor to rotate for one revolution. The input terminal of the central processing unit receives image signals to analyze and process the collected three-sided images. The display unit connected to the central processing unit displays the collected images and processing results. Products are rejected. The feeding trough unit has two outlets for electrolytic capacitors, one is the outlet for qualified electrolytic capacitors, and the other is the outlet for unqualified products. A manipulator is provided to transfer the unqualified products judged by the defendant in the transparent trough of the feeding chute unit to another box .

The central processing unit is used to analyze and process the collected three-sided images, the display unit displays the collected images and the processing results, and the unqualified product rejecting unit rejects unqualified products that are detected to have defective appearance and packaging.

As shown in Figure 2, an image detection method for the appearance and packaging defects of electrolytic capacitors uses the three-direction images obtained from the image acquisition of the upper surface, the image acquisition of the side surface and the image acquisition of the lower surface for image processing to identify unqualified products with defective appearance. It is characterized in that the upper and lower surface images are respectively subjected to color image grayscale, image denoising (linear filter method, median filter method, Wiener filter method, Fourier transform and wavelet transform), image enhancement (such as histogram, low pass filter, histogram equalization, etc.), region segmentation, image normalization (normalized grayscale combination method), defect detection. Color image grayscale, image denoising, image enhancement, image stitching, region segmentation, image normalization, and defect detection are performed on the side image respectively. The grayscale of the color image is used to convert the color image acquired by the camera into a grayscale image, the image denoising is used to remove noise from the grayscale image, and the image enhancement is used to enhance the contrast of the image , the region segmentation is used to segment the region to be detected, the image normalization is used to perform normalization processing such as rotation, cropping and scaling on the image to be detected, and the defect detection is used to normalize the image to be detected Defect identification is performed on the normalized image and the existing standard normalized image template, and the image mosaic is used to mosaic the capacitive acquisition image that rotates one revolution into a complete side image. Specific steps are as follows:

Step 001. As soon as the photoelectric switch detects the capacitance, it triggers the camera on the upper surface image acquisition unit to capture the image;

Step 002. Convert the upper surface color image into a grayscale image;

Step 003. Filter the upper surface grayscale image to remove image noise;

Step 004. Perform image enhancement on the denoised upper surface grayscale image to enhance the contrast of the image;

Step 005. Perform region segmentation on the enhanced upper surface image, specifically including the following steps:

Step 0051. Utilize the sobel edge detection operator to perform edge detection on the enhanced upper surface image;

Step 0052. Utilize the hough transform on the upper surface edge image to detect the appearance edge circle curve on the upper surface of the capacitor;

Step 0053. Mark the area where the circular curve on the appearance edge of the connected capacitor is located, and obtain the segmented upper surface area to be detected;

Step 006. In particular, if there is character identification information on the upper surface packaging, character identification detection is also required, which is characterized in that it includes the following steps:

Step 0061. Perform horizontal or vertical grayscale projection on the segmented upper surface area to be detected, and segment the area where the specification information of the upper surface appearance of the capacitor is located;

Step 0062. Rotate the upper surface to-be-detected area obtained in step 0053 to the vertical direction by using the angle of the line where the upper-left corner and the lower-left corner pixel in the image of the upper-surface specification information to-be-detected area deviates from the vertical direction; wherein the upper-left corner pixel is from From left to right, the first pixel found from top to bottom, the bottom left pixel is the first pixel found from left to right, from bottom to top.

Step 007. Normalize the segmented region image, specifically including the following steps:

Step 0071. The image of the area to be detected on the upper surface rotated to the vertical direction is cropped into a square image with 5 pixels in the upper, lower, left, and right sides of the circle edge;

Step 0072. Scale the cropped image of the area to be detected on the upper surface to a specified calibration image of a specified pixel size (such as 160×160);

Step 008. Carry out an XOR operation on the normalized upper surface to-be-detected area image and the standard upper surface template image in the image database (the acquisition process is also through steps 001-007). If the calculation result exceeds the threshold, it is determined that the upper surface is not qualified.

Step 009. The second photoelectric switch detects the capacitance, and triggers the camera of the image acquisition unit on the lower surface to capture the image;

Step 0010. Convert the lower surface color image into a grayscale image;

Step 0011. Filtering the lower surface grayscale image to remove image noise;

Step 0012. Perform image enhancement on the denoised lower surface grayscale image to enhance the contrast of the image;

Step 0013. Perform region segmentation on the enhanced lower surface image, specifically including the following steps:

Step 00131. Utilize the sobel edge detection operator to perform edge detection on the enhanced lower surface image;

Step 00132. Utilize the hough transform on the lower surface edge image to detect the appearance edge circle curve of the capacitor lower surface;

Step 00133. Mark the area where the circular curve of the appearance edge of the connected capacitor is located, and obtain the segmented lower surface area to be detected;

Step 0014. In particular, if there is character identification information on the lower surface packaging, character identification detection is also required, which is characterized in that it includes the following steps:

Step 00141. Perform horizontal or vertical grayscale projection on the segmented lower surface area to be detected, and segment the area where the specification information of the lower surface appearance of the capacitor is located;

Step 00142. Rotate the area to be detected on the lower surface obtained in step 00133 to the vertical direction by using the angle of the line where the upper left corner and the lower left corner pixel in the image of the area to be detected in the specification information of the lower surface deviates from the vertical direction; wherein the upper left corner pixel is from From left to right, the first pixel found from top to bottom, the bottom left pixel is the first pixel found from left to right, from bottom to top.

Step 0015. Normalize the segmented region image, specifically including the following steps:

Step 00151. Cutting the image of the area to be detected on the lower surface rotated to the vertical direction into a square image with 5 pixels in the top, bottom, left, and right sides of the circle;

Step 00152. Scale the cropped image of the area to be detected on the lower surface to a specified calibration image of a specified pixel size (such as 160×160);

Step 0016. Carry out XOR operation with the normalized image of the area to be detected on the lower surface and the standard lower surface template image in the image database (the acquisition process is also through steps 009-0015), and the judgment that the result of the operation exceeds the threshold is that the lower surface is not qualified.

Step 0017. The photoelectric switch 3 detects the capacitor, triggers the side surface image acquisition unit camera to capture the image, and simultaneously triggers the rotating plate on the feeding trough to drive the capacitor to rotate for one revolution;

Step 0018. Convert a plurality of side surface color images into grayscale images;

Step 0019. Filtering and removing image noise on multiple side surface grayscale images respectively;

Step 0020. Perform image enhancement on the multiple side surface grayscale images after denoising to enhance the contrast of the image;

Step 0021. Perform image mosaic on the enhanced side surface images, specifically including the following steps:

Step 00211. Extract feature points from the left and right ends of the enhanced side images;

Step 00212. Matching the feature points extracted from the left and right ends;

Step 00213. Perform image fusion according to the matching points to obtain the entire side image after splicing.

Step 0022. Perform horizontal or vertical grayscale projection on the entire side image obtained by splicing, and segment the area where the specification information of the surface appearance of the capacitor side is located;

Step 0023. Rotate the area to be detected on the side surface to the vertical direction by using the angle at which the straight lines of the upper left corner and lower left corner pixels in the image of the area to be detected in the side surface specification information deviate from the vertical direction; wherein the upper left corner pixels are from left to right, The first pixel found from top to bottom, the bottom left pixel is the first pixel found from left to right and bottom to top.

Step 0024. Normalize the segmented region image, specifically including the following steps:

Step 00241. Crop the image of the area to be detected on the side surface rotated to the vertical direction into a square image of 5 pixels from the edge, up, down, left, and right;

Step 00242. Scale the cropped image of the area to be detected on the side surface to a specified calibration image with a specified size of pixels (such as 160×160);

Step 0025. Carry out XOR operation with the normalized side surface to-be-detected area image and the standard side surface template image in the image database (its acquisition process is also through steps 0017-0024), and the judgment that the operation result exceeds the threshold is that the lower surface is not qualified.

Furthermore, the present invention designs a system and method for detecting defects in the appearance and packaging of electrolytic capacitors. The standard upper surface template image is produced by qualified products through steps 001 to 007, and the standard lower surface template image is produced by qualified products. Use steps 009 to 0015 for processing.

As a preferred technical solution, the direction of the long capacitor pin detected by the photoelectric switch can be combined to determine the direction of the positive symbol in the appearance image.

As a preferred technical solution, the display unit can display the whole visual inspection process.

As an optimal technical solution, it is possible to add a device for displaying unit display functions of unqualified products according to defect categories and removing different defect categories according to needs.

An image detection method for appearance packaging defects of electrolytic capacitors designed by the present invention, as a preferred technical solution, selects the number of images for image splicing according to the field of view of the camera of the side surface acquisition unit.

An image detection system and detection method for electrolytic capacitor appearance packaging defects designed by the present invention can replace the cumbersome detection operation process of traditional manual visual inspection, and automatically determine whether there is a defect in the appearance of electrolytic capacitor packaging, and the display unit can display the entire judgment cycle in real time In fact, it can also be classified and eliminated according to defect categories (such as upper surface defects, side positive and negative electrode defects, side model information defects, and lower surface defects). Compared with the traditional manual detection operation, the system and method designed by the present invention have the advantages of simple structure, high detection efficiency, high discrimination rate, fast speed, high reliability, easy implementation, low cost and strong applicability. Typical product packaging quality such as: the edges of the plastic edging are not aligned with the edges of the capacitor, and the edges of the plastic edging are broken.

In summary, by establishing and implementing the visual imaging device-based electrolytic capacitor appearance packaging defect image detection system and method designed by the present invention, it is possible to realize whether the upper and lower surface packaging edges, side positive and negative pole marks, and side model information on the packaging of capacitor products are High-efficiency and effective automatic detection of defects or errors has broad market application prospects and economic value.

The embodiments of the present invention have been described in detail above in conjunction with the accompanying drawings, but the present invention is not limited to the above embodiments, and can also be made without departing from the gist of the present invention within the scope of knowledge possessed by those of ordinary skill in the art. Variations.

Claims (6)

1. a kind of electrolytic capacitor appearance packaging defect image detection method, it is characterized in that being based on following detecting system, detecting system Including feed well unit, light source Unit one, upper surface image collecting unit, optoelectronic switch Unit one, light source Unit two, lower surface Image acquisition units, light source Unit three, optoelectronic switch Unit three, side surface image acquisition units, central processor unit, display Unit and substandard product culling unit；The feed well unit presses tooth to capacitance into between-line spacing feeding by vibrating disk, vibrates The outlet of disk is feed well, and vibrating disk is a concavity disk with helix, and feed well is that the transmission that two rectangular slabs limit is saturating Flume, transmission transparent cell centre are equipped with electrolytic capacitor detection station, and electrolytic capacitor pin is inserted in transparent cell lower end；It is set on transparent cell There are three station, light source Unit one is located at the surface of the upper surface image collecting unit of station one, is acquired for upper surface image Unit provides light, and upper surface image collecting unit carries out the upper surface of capacitance after optoelectronic switch Unit one detects capacitance Appearance images acquire；Light source Unit two is located at the underface of the lower surface image collecting unit positioned at station two, is lower surface figure As collecting unit offer light, lower surface image collecting unit is after optoelectronic switch Unit two detects electrolytic capacitor to capacitance Lower surface carries out appearance images acquisition；Light source Unit three is located at the side of the side surface image acquisition units of station three, is side table Face image acquisition units provide light, and side surface image acquisition units are after optoelectronic switch Unit three detects capacitance to capacitance Side surface carries out appearance images continuous acquisition, and station San Tong method is rotary work station, rotary work station（1-3）Drive pin that capacitance is made to revolve It circles, optoelectronic switch Unit three, which detects, triggers rotation when capacitance enters station three, i.e., the stepping electricity on feed well on station Machine swivel plate band dynamic condenser rotates a circle；Central processor unit input terminal receives picture signal for the three-view drawing picture to acquisition Analyzing processing is carried out, the display unit display acquisition image and handling result, substandard product of central processor unit connection pick Except unit will detect that the defective substandard product of appearance packaging is rejected；

In image detecting method, the upper and lower surface image of acquisition is subjected to coloured image gray processing, image denoising, image respectively and is increased By force, region segmentation, image normalization, defects detection；The side image of acquisition carries out coloured image gray processing respectively, image is gone It makes an uproar, image enhancement, image mosaic, region segmentation, image normalization, defects detection；The coloured image gray processing is used for will The coloured image that video camera obtains is converted into gray level image, and the image denoising is used for carrying out noise remove to gray level image, The image enhancement is used for carrying out contrast enhancing to image, and the region segmentation is used for being partitioned into area to be tested, institute The image normalization stated is used for being rotated, cut and scaled normalized, the defects detection to image to be detected For carrying out defect recognition, the image mosaic to normalized image to be detected and existing standard normalized image template For acquiring image mosaic into the complete side image of a width to the capacitance to rotate a circle；

It is as follows：

Step 001. optoelectronic switch one detects capacitance, triggers upper surface image collecting unit camera captures images；

Upper surface coloured image is converted into gray level image by step 002.；

Step 003. is filtered removal picture noise to upper surface gray level image；

Step 004. carries out image enhancement to enhance the contrast of image to the upper surface gray level image after denoising；

Step 005. carries out region segmentation to enhanced upper surface image, specifically includes below step：

Step 0051. carries out edge detection to enhanced upper surface image using sobel edge detection operators；

Step 0052. utilizes hough change detection capacitances upper surface appearance boundary curve to upper surface image edge；

Step 0053. label connection capacitance appearance boundary curve region, the upper surface area to be tested divided；

If having character mark information in the packaging of step 006. upper surface, also needs to carry out character mark detection, include the following steps：

The upper surface area to be tested that step 0061. obtains segmentation does horizontally or vertically Gray Projection, is partitioned into capacitance upper table The specification information region of face appearance；

Step 0062. is inclined using straight line where the upper left corner in the specification information area to be tested image of upper surface and lower left corner pixel Angle from vertical direction rotates to vertical direction the upper surface area to be tested that step 0053 obtains；The wherein upper left corner Pixel is that from left to right, first pixel found from top to bottom, lower left corner pixel is from left to right, to find from bottom to up First pixel；

Cut zone image is normalized in step 007., specifically includes below step：

Step 0071. is to rotating to the upper surface area to be tested image cropping of vertical direction into edge each 5 pictures up and down The square-shaped image of element；

Step 0072. to the upper surface area to be tested image scaling after cutting extremelyPixel calibrates image；

Step 008. is by the standard upper surface Prototype drawing in normalized upper surface area to be tested image and image data base Picture, acquisition process are also through step 001-007, carry out XOR operation, and operation result is more than that the judgement of threshold value does not conform to for upper surface Lattice；

Step 009. optoelectronic switch two detects capacitance, triggers lower surface image collecting unit camera captures images；

Lower surface coloured image is converted into gray level image by step 0010.；

Step 0011. is filtered removal picture noise to lower surface gray level image；

Step 0012. carries out image enhancement to enhance the contrast of image to the lower surface gray level image after denoising；

Step 0013. carries out region segmentation to enhanced lower surface image, specifically includes below step：

Step 00131. carries out edge detection to enhanced lower surface image using sobel edge detection operators；

Step 00132. utilizes hough change detection capacitances lower surface appearance boundary curve to lower surface image edge；

Step 00133. label connection capacitance appearance boundary curve region, the lower surface area to be tested divided；

Step 0014. particularly, if lower surface packaging on have character mark information, also need carry out character mark detection, including with Lower step：

The lower surface area to be tested that step 00141. obtains segmentation does horizontally or vertically Gray Projection, is partitioned under capacitance The specification information region of appearance；

Step 00142. is inclined using straight line where the upper left corner in the specification information area to be tested image of lower surface and lower left corner pixel Angle from vertical direction rotates to vertical direction the lower surface area to be tested that step 00133 obtains；Wherein upper left Angle pixel is that from left to right, first pixel found from top to bottom, lower left corner pixel is from left to right, to find from bottom to up First pixel；

Cut zone image is normalized in step 0015., specifically includes below step：

Step 00151. is to rotating to the lower surface area to be tested image cropping of vertical direction into edge each 5 pictures up and down The square-shaped image of element；

Step 00152. to the lower surface area to be tested image scaling after cutting extremelyPixel calibrates image；

Step 0016. is by the standard lower surface Prototype drawing in normalized lower surface area to be tested image and image data base As carrying out XOR operation, acquisition process is also through step 009-0015；Operation result is more than that the judgement of threshold value does not conform to for lower surface Lattice；

Step 0017. optoelectronic switch three detects capacitance, triggers side surface image acquisition units camera captures images, touches simultaneously Swivel plate band dynamic condenser on hair feed well rotates a circle；

Multiple side surface coloured images are converted into gray level image by step 0018.；

Step 0019. is filtered removal picture noise respectively to multiple side surface gray level images；

Step 0020. carries out image enhancement to enhance the contrast of image respectively to multiple side surface gray level images after denoising；

Step 0021. carries out image mosaic to enhanced multiple side surface images, specifically includes following steps：

Step 00211. extracts characteristic point respectively in the left and right ends of enhanced multiple side images；

Step 00212. carries out homotopy mapping to the characteristic point that left and right ends extract；

Step 00213. carries out image co-registration according to match point, obtains spliced whole picture side image；

The whole picture side image that step 0022. obtains splicing does horizontally or vertically Gray Projection, is partitioned into outside capacitance side surface The specification information region of sight；

Step 0023. is inclined using straight line where the upper left corner in the specification information area to be tested image of side surface and lower left corner pixel Angle contralateral surface area to be tested from vertical direction is rotated to vertical direction；Wherein top left corner pixel be from a left side to The right side, first pixel found from top to bottom, lower left corner pixel are first pixels found from bottom to up from left to right Point；

Cut zone image is normalized in step 0024., specifically includes below step：

Step 00241. is to rotating the side surface area to be tested image cropping to vertical direction into each 5 up and down away from edge The square-shaped image of pixel；

Step 00242. calibrates figure to the regulation of the side surface area to be tested image scaling after cutting to size determined pixel point Picture；

Step 0025. is by the standard side surface Prototype drawing in normalized side surface area to be tested image and image data base As carrying out XOR operation, acquisition process is also through step 0017-0024, and operation result is more than that the judgement of threshold value does not conform to for lower surface Lattice.

2. a kind of electrolytic capacitor appearance packaging defect image detection method according to claim 1, it is characterised in that：The mark Quasi- upper surface template image is to handle to generate using step 001 ~ 007 by qualified products to get, and standard lower surface template image is It handles to generate using step 009 ~ 0015 by qualified products and get, standard Side shuttering image is to use step by qualified products 0017 ~ 0024 processing, which generates, to be got.

3. a kind of electrolytic capacitor appearance packaging defect image detection method according to claim 1, it is characterised in that：The light The long pin direction of capacitance that electric switch detects, which can combine, is used for differentiating positive symbol direction in appearance images.

4. a kind of electrolytic capacitor appearance packaging defect image detection method according to claim 1, it is characterised in that：Described Display unit whole can show vision inspection process.

5. a kind of electrolytic capacitor appearance packaging defect image detection method according to claim 1, it is characterised in that：It can root According to the device for needing to add substandard product by defect classification display unit display function and the rejecting of different defect classifications.

6. a kind of electrolytic capacitor appearance packaging defect image detection method according to claim 1, it is characterised in that：According to side The visual field of surface image collecting unit camera chooses the picture number for carrying out image mosaic.