Detection Method of Longitudinal Tear for Mine Conveyer Belt Based on Improved Gaussian Mixture Model

An online detection method for longitudinal tear of mine conveyor belt is proposed, and the method combines infrared image features with improved Gaussian Mixture Model(GMM). An adaptive hybrid median filtering technique is designed. In view of the error prone initialization of Gaussian Mixture Model, the problem is improved by using the weighted alternative fuzzy C-means, the infrared image feature parameters are used as eigenvectors of improved GMM clustering for cluster analysis to realize the longitudinal tear identification of mine conveyor belt. The experimental results show that the correct recognition rate for the longitudinal tear detection of conveyor belt can reach 99% with this method.