The number of domestic companies about machine vision detection is bigger and bigger, simple detection projects are gradually less and less. The inspection projects left are gradually more complicated and tedious. It is a challenge for visual inspection companies and is also a chance of self-promotion. We'll overcome difficult and complex projects like this .
At present, defect detection is the most difficult in the industry of machine vision detection. All kinds of scratches, bruises, stains and cracks need to be well reflected in the imaging end to facilitate the detection of back-end algorithm.
Goldfinger is the essential part of the memory chips, SD card, U disks and other storage devices, which determines the quality of the entire product. The goldfinger surface of the product has the gold plating with a certain thickness. If there is any damage on the surface, the brass of the lower level is easy to cause poor contact oxidation to become defective goods cost loss. So in each link, manufacturers arranges artificial detections or use other auxiliary tools. The purpose is to put an end to bad products, but it often cannot achieve the desired effect and still can appear a lot of bad products.
After knowing the requirements of the manufacturer, we analyzed the processing technology and process which were prone to defects. Then we designed a set of detection system for defective products of goldfinger. Leak copper, scratches, stains, defects such as solder point for testing, the system uses high resolution camera with the GIGE gigabit, cooperating with the company’s self-developed single band high-resolution telecentric lens. The system has good stability, low detection accuracy and high error detection rate, presenting the goldfinger’s defects on each corner perfectly. Through image
By using the ring light source and adjusting the appropriate height of the light source, the copper leakage, scratches, stains, tin spots and other defects on the surface of goldfinger can be photographed. Due to the different forms of defects, the feedback image information is different, but the important thing is that the background can be clearly distinguished from the defects. The scheme can be used to detect metal plane defects with relatively flat surface.