Research On Defect Detection Method Of Wire Inside Optical Device Based On Machine Vision

As an important part of the optical transceiver module,the production quality of the optical device has a great influence on the working time of the optical module.The wire inside the optical device is connected to the chip and the pad to transmit electrical signals.Therefore,the welding quality of the wire inside the optical device is related to the production quality of the optical device.At present,companies still use the traditional manual visual inspection method to detect the welding quality of the wire inside the optical device,which is inefficient and the inspection results are affected by people’s subjective factors.In order to improve the detection speed of the welding quality of the gold wire inside the optical device,thereby improving the production efficiency of the optical device,this paper takes laser diode chip on carrier as the research object,and then proposes a defect detection method for the wire inside optical devices based on machine vision.The method realizes the automatic detection of the welding quality of the wire inside the optical device and the classification of the defect type.The main research contents of this paper are as follows:(1)The original image of LDCOC is processed to facilitate later extraction of wires and defect identification.Firstly,the original color image is converted into a grayscale image.And then,different image filtering methods are used to reduce noise in images.Finally,aiming at the LDCOC deflection problem caused by the vibration of the conveyor belt,this paper designs an automatic correction algorithm to realize the automatic rotation of the LDCOC image and the operation of removing black edges.(2)Aiming at the problem that the reflective area of the wire interferes with the later wire extraction during LDCOC image acquisition,an algorithm for removing the reflective area of the wire is designed in this paper.When the LDCOC image is collected,the area with high curvature of the wire welding is prone to reflection problems,and the reflection area will interfere with the extraction of the internal wire image.Therefore,the normalized correlation matching method is used to determine the chip area.And then,according to the characteristics of the reflective area as a closed small area,an eight-neighbor replacement algorithm is proposed to remove the reflective area,which provides a basis for subsequent internal wire defect detection.(3)This paper analyzes the types of wire bonding defects in the LDCOC image after removing the reflective area,and then proposes corresponding welding defect detection methods respectively.By analyzing the characteristics of the Welding point of the pads in the chip,the welding of the wires and the welding point in the pad welding area,different detection schemes are used to achieve the detection target of gold wire defects.First,for the defect detection of the welding point of the chip pads,the Hough Transform circular detection method is used to obtain the pad area,and then according to the pixel value difference of the pad area,it is judged whether there are welding point on the pad and whether the welding point exceed the pad area.Then,for the detection of the wire welding defects,the method of regional growth is used to extract the wire,and the distance between the two welding points is determined according to the calculated maximum circumscribed rectangle of the welding wire,so as to determine whether the wire is disconnected.Finally,according to whether there is a welding point problem in the pad area of the pad,a method is proposed to compare the coordinates of the welding point at one end of the wire with the coordinates of the pad welding area,so as to judge whether the identified welding point of the wire exceeds the pad welding area.The experimental results show that defect detection scheme for the wires inside the optical device proposed in this paper can automatically identify four defect types in LDCOC: no welding wire,wrong welding point position in the pad,disconnected welding wire and wrong welding point position in the welding area of spacer.It has strong theoretical research significance and practical application value.