Research On Image-based Measurement Method For Printed Circuit Board Parameters

Pads and lines are important parts of printed circuit board(PCB), the quality ofthem affects the performance of circuit board directly. Rapid, accurate and stablemeasurement for the parameters of pads and lines is a critical step in the productionprocess, the purpose is to determine whether the parameters conform to therequirements for the design of the circuit board. In this paper, we mainly research themeasurement method for the parameters of PCB, the reason is that traditional methodhas the low degree of automation, low precision and low efficiency. Specific researchcontents and the results obtained are as follows:(1) Constructing the image-based measurement system for the parameters of PCB.According to the selection principle for cameras, image acquisition card, lenses, lightingequipment and other components, we construct the hardware part for the measurementsystem. The overall software system is designed on the basis of functional needs, andthe functional modules are divided.(2) Study the pretreatment method for the PCB image. Correct the PCB image tobe measured by using coordinate transformation to the corresponding points，on thebasis of analyzing images of non-linear distortion model and image distortion correctionprinciple, aim for avoiding the influence of image distortion. Remove the noise on PCBby using median filter can protect the image edge preferably. Otsu algorithm is appliedto segment the PCB image automatically, aim for separating pads and lines frombackground. According to the different gray value between pads and background, weremove the teardrop of pad by scanning pixels on the circumference to reduce theimpact for measurement precision from teardrop. Obtain clear and continuous edges ofpads and lines for subsequent image processing by using Robert algorithm.(3) Research the image-based measurement method for the parameters of printedcircuit board. Firstly, we extract the edge point set of pad and line with point-by-pointscanning method; Secondly, remove the abnormal edge points by using the least-squaresfitting circle and least squares linear fitting method respectively; Thirdly, locate theoriginal edge points in sub-pixel level with Zernike moment, and fit circular andlinear equations with the sub-pixel edge points. Finally, the optical ruler is applied to calculate the actual distance between neighbor pixels. In order to validate theeffectiveness of proposed measurement method, we measure the parameters for pad andline of real PCB many times. The experimental results show that the methodcan measure the parameters of pads and lines with high efficiency, high precision andhigh stability.