Visual tracking method of shearer based on compressive sensing

For problems of low illumination intensity, uneven illumination and high coal dust concentration in working face, a visual tracking method of shearer based on compressive sensing was proposed. The image is normalized by use of rectangular filter firstly to get feature vectors. Then compressed Haar-like feature vectors of target samples and background samples are gotten according to compressive sensing theory for building target model and training naive Bayes classifier. The target image and background image are identified by the naive Bayes classifier finally, so as to realize dynamic tracking of shearer. The experimental result shows that the method can track shearer effectively when the shearer is moving or covered in environment of uneven and varied illumination, and average tracking frame rate is 22 frames per second.