Research on Machine Vision Intelligent Detection Algorithm Based on Sparse Representation and Markov-Bayesian Reasoning
作者单位：School of Electrical Engineering Nantong University Nantong Research Institute for Advanced Communication Technologies
会议名称：《第37届中国控制会议》
会议日期：2018年
学科分类：12[管理学] 1201[管理学-管理科学与工程(可授管理学、工学学位)] 081104[工学-模式识别与智能系统] 08[工学] 080203[工学-机械设计及理论] 0835[工学-软件工程] 0802[工学-机械工程] 0811[工学-控制科学与工程] 0812[工学-计算机科学与技术（可授工学、理学学位）]
基 金：financiallysupported by Nantong University-Nantong Joint Research Center for Intelligent Information Technology under Grant No.KFKT2016A08 by Nantong Research Program of Application Foundation under Grant No. GY12016022
关 键 词：light spot sparse representation dictionary learning markov-bayesian
摘 要：At present, the product testing on high-speed assembly line mainly depends on manual testing. Automatic Depression cigarettes detection system is a machine vision method instead of manual testing. The utility model can overcome the defects of easy inspection, high false detection and slow speed of human inspection, and meets the development trend of industrial automation production and the continuous improvement of people’s requirements on the quality of cigarettes. In order to quickly and accurately detect complex detonation beads with background texture and improve the computational complexity of traditional algorithms, this paper proposes a dictionary learning algorithm based on sparse representation of cigarettes detonation to indirectly detect the light spots in the image by reducing the dimensionality At the same time, the probability of flare distribution in the shot bead can be calculated by using the algorithm of Markov-Bayesian inference, which can be effectively used in the shot detection. In this experiment, a high-performance industrial computer and a self-developed bursting ball detection algorithm were used to detect the explosive beads on the assembly line, so that the bursting of the added cigarettes could be quickly and accurately detected.