关键词: 异纤检测, 光源优化设计, 电荷耦合元件, 模糊聚类分析

Abstract:

In order to allow Charge-coupled Device CCD to accurately collect and deal with the different fiber image and detect the multiple types of different fibers, a new method based on fuzzy clustering neural network is proposed. Detailed describe the working principle of this kind of light source system, through the analysis of the relationship between CCD imaging and the incident light energy to determine the number of test of the light source; establish of exposure as a function of the amount of the CCD target surface, analyzes the optimum light source position detection; finally, the background image for the CCD plate light distribution and average gray level through the parametric equation of image analysis, by means of fuzzy clustering analysis, considering the input values of all the information build multi type of light source, fuzzy clustering neural network, optimization design of the light source and the number of distance. The design result shows that the best detection position is different fiber in position 1 / 2 of the tube, in the number of light source system for 10, distance on both sides of the light source is 3mm, the neural network convergence error reached the expected value, and the foreign fiber detection rate reaches 94.79 %, meeting the requirements of actual production.

Key words: foreign fiber detection, optimization design of light source, charge-coupled device, fuzzy cluster analysis