School of Mechanical Electronic and Information Engineering,China University of Mining and Technology (Beijing),Beijing　 100083,China

The mine intelligent monitoring is an important guarantee for the realization of automatic mining and visual operation in few people or unmanned working face. To address the problems of low resolution in the existing captured videos,the image blur with noise and long processing time by using the conventional methods of Nyquist sampling and image compressing for mine monitoring images in the mine environment where the images are easy to be influenced by the noise and spray dust,on the basis of the compressed sensing and sparse reconstruction theory,a novel monitoring images improved algorithm based on sparsity adaptive matching pursuit ( SAMP) is proposed. Firstly,by establishing the model of the block compressed sensing in images,the proposed method adopts sparse random measurement matrix to compress sensing block images and to obtain observations. Then,DFT as sparse basis for signal sparse representation is used. Finally,an improved adaptive matching pursuit algorithm is employed to decode the images. The results indi- cate that the method proposed in this paper shows the superiority in comparing with other algorithms,effectively im- proves the speed of image acquisition and compression,and enhances decoding quality in the reconstruction process,as well as has a strong anti-noise performance and robustness. This method is helpful to improve the image definition and real-time processing performance for mine monitoring system.

国家重点研发计划重点专项资助项目(2016YFC0801800)；国家自然科学基金资助项目(51674269)；中央高校基本科研业务基金资助项目(2014YJ01)；

Zhang Fan,Yan Xiuxiu,Li Yajie. A novel image reconstruction method of mine intelligent surveillance based on adaptive sparse representa- tion[J]. Journal of China Coal Society,2017,42(5):1346-1354.