Machine vision based LCD spot flaw detection method and system

Abstract

The disclosed detect method for LED display stain based on machine vision comprises: acquiring some images of target display to filter noise and one polynomial curves that can show the data trend; subtracting the image by the polynomial curve to separate the pixel over threshold as the suspected stain target; selecting the contrast grade, area, edge parameter, position, gray scale evenness, and shape of the suspected target as input variables; selecting stain defect grade as output variable; dividing the input and output variables into some fuzzy subsets with membership grade function.

Description

technical field

[0001] The invention relates to a spot defect detection system of a liquid crystal display, in particular to a machine vision-based spot defect detection system. Background technique

[0002] The production process of liquid crystal displays is very complicated, and although most of the processes are completed in a clean room, some visual defects will inevitably appear. There are many types of visual defects in liquid crystal displays, which can generally be divided into three categories according to the area and shape of the defects: point defects, line defects and surface defects. Among them, surface defects can be divided into two types: block defects and spot defects (called mura in English, which comes from Japanese, meaning dirt and spots, and is the most common surface defect). Point defects, line defects, and block defects are mainly caused by electrical causes such as short circuits, open circuits, and bad transistors in thin-film transistor (TFT) a...

Images