High-density flexible substrate appearance defect detection system and method based on depth learning

Abstract

The invention discloses a high-density flexible substrate appearance defect detection system and a method based on depth learning, the method comprises a hardware platform and a software detection platform. The detecting method comprises the following steps: FICS images containing different defects is collected as training samples; pre-processing on the sample image is carried out, including the unified standard size and manual marking of the defect position and the category in the sample; a sample image is input into a depth learning model based on an improved YOLO convolution neural networkto carry out training to obtain model parameters which are output as defect positions and categories; the collected image size is standardized and input into a trained depth learning model for detection, and the defect position and the category information in the acquired image are obtained. The method can achieve the rapid positioning and the type identification of the appearance defects of the high-density flexible substrate, and solves the problem that the traditional defect detection system and the method are low in speed and is difficult to achieve high-density FICS appearance defects.

Description

technical field [0001] The invention belongs to the technical field of machine vision surface defect detection, and in particular relates to a high-density flexible substrate appearance defect detection system and method based on deep learning. Background technique [0002] Flexible Integrated Circuit Substrate (FICS for short) is a high-density flexible printed circuit board that can be used as an IC packaging substrate. In the production process of high-density FICS, due to the precision of process control, it is inevitable to have defects in appearance. Rapid positioning and type identification of various appearance defects of FICS through high-precision visual inspection is the key to quality control in the high-density FICS manufacturing process. [0003] At present, manufacturers mainly use manual visual inspection with a magnifying glass to achieve high-density FICS appearance defect detection. This method has low detection efficiency, consumes a large amount of lab...

AI Technical Summary

Problems solved by technology

[0005] In view of the low efficiency of manual visual inspection and traditional appearance defect detection systems and methods in the manufacturing process of high-density flexible substrates, and the inability to detect multiple defects at the same time, the present invention provides a high-density flexible substrate appearance defect detection system and method based on deep learning. method to achieve rapid location and type identification of appearance defects on high-density flexible substrates

Images