Detection and Identification of Defects in Transparent Film

• DOI: 10.14257/ijfgcn.2015.8.4.9
• Publication Date: 2015-12-12T02:57:01Z
• Authors: Bao -yuan Chen; Zheng Chao; Zhong-xiang Sun; Xiao-yang Yu
• Keywords: Identification

Biaxially oriented polyester film (BOPET) defect is an important factor affecting the quality of film.In view identification defects in conventional production process, this mathod resulted inaccurate.and low labor efficiency and machine vision recognition on specific defect.This paper presents a LVQ neural network-based BOPET detection methods.In algorithm, images were processed outlines membrane obtained.Through extracting aspect ration, circularity, complexity elongation , projection histogram central moment so on, characteristic values defects, which from image after processing, then input to system based network that had been trained, order achieve identification, classification localization.Through study features extracted some quantities as character network, training value into learning prediction purpose, was designed.The experiments show that, proposed method can meet requirements analysis air transparent film.