superpixel-based copper surface defect detection method for a flexible IC packaging substrate

Abstract

The invention discloses a superpixel-based copper surface defect detection method for a flexible IC packaging substrate, and the method comprises the steps: carrying out the filtering and preliminarythreshold segmentation of an original color image, and obtaining a superpixel segmentation image through employing a simple linear iterative clustering method; Selecting a superpixel of the copper surface area according to the brightness characteristic, extracting the characteristic of the superpixel of the copper surface area by combining the global characteristic of the image, calculating the brightness difference degree of pixel points in the superpixel, finding out a hole in the copper surface, and oxidizing a defect area; Secondly, finding out a superpixel located at the joint of the copper surface and the background according to the gradient and neighborhood information of the pixel, obtaining boundary pixel points located at the edge of the copper surface, obtaining a continuous copper surface line by scanning the neighborhood of the boundary pixel points, and performing feature extraction and analysis on the edge pixel points on the copper surface line to find out the line sawtooth defect. Defects such as holes, oxidation and circuit sawteeth of a copper surface can be accurately positioned.

Description

technical field [0001] The invention relates to the application field of image processing technology, in particular to a method for detecting defects on a copper surface of a superpixel-based flexible IC packaging substrate. Background technique [0002] Flexible IC packaging substrates are generally composed of polyimide, polyester film and other polymer materials, which are light, thin and bendable, and have a wide range of application markets in the circuit board manufacturing industry in the fields of communications, electronics, and aviation. With the development of technology, flexible IC packaging substrates gradually tend to be high-density and high-precision, and the quality requirements are becoming more and more stringent. The manufacturing process integrates knowledge of chemistry, physics, polymers, optics, etc., and the technology is complex. , so it is easy to generate all kinds of quality defects, thereby affecting the quality of the substrate. The manual vi...

AI Technical Summary

Problems solved by technology

This method makes full use of the information of the image, but usually the feature dimension is high, the amount of calculation is large, the algorithm framework is complex, and it will involve the determination of many parameters, which is not conducive to the promotion and application in the actual production line

The edge detection method generally extracts the edge or contour of the image through an edge detection operator or a specific line transformation algorithm, calculates the corresponding statistical information, and combines color or geometric features to obtain a decision strategy, but this method usually involves binarization and segmentation of images and line fitting , the poor selection of the segmentation threshold will affect the segmentation effect, and even blur the image details, resulting in low detection accuracy, long time-consuming and complicated calculation in the line fitting process

Images