Research And Application Of Defect Detection Method Based On Image Processing Chip Fixed Resistor

With the rapid popularization of Surface Mount Technology, most electronic products haveadopted SMT. The application of SMD components improves the production efficiency and reducesthe production costs of electronic products. Chip resistors which are essential components ofelectronic products, it’s quality is one of the key factors of electronic products’quality and pass rate.Thus, the study of using machine vision to achieve fast an accurate detection of chip resistors issignificant needed.Image filtering techniques and image enhancement techniques is introduced, and we compareMedian Filtering, Gaussian Filtering and Mean Filtering on chip resistors image filtering effect. Inthis paper we improve Median Filtering technique by the factors of chip resistors image for whichprovide higher SNR.We compare the factors of template matching, fast template matching and dynamic templatematching, and improve the performance of dynamic template matching by chip resistors imagefactors. We detect the defects of chip resistors surface by traditional template matching method anddynamic template matching method respectively, the correct detection is89.6%and94.3%.We design feature extraction method for chip resistors surface defects which based on PCAand ICA. This operation is not only largely compress the image date but also obtain thecharacteristic coefficient.We research the basic theory of support vector machine，we adoptone-to-many maximum response value to classify the chip resistors surface defects. The results ofthe experiment show that the correct classification rate is80%.