In view of problems of current video monitoring system on fully-mechanized mining face such as occupying large network bandwidth, incomplete video storage, unobtrusive cutting pictures of the shearer, and uneven video stitching pictures, optimization design was carried out from aspects of system hardware and software, video compression and stitching algorithms, etc. In terms of hardware, hard disk video recorders are introduced to reduce network bandwidth occupancy rate and solve the problem of video transmission jam; the combination of local storage and remote storage effectively solved the problem of video storage loss. In terms of software, based on the principle of highlighting the key points and partial zooming in, combination of real-time video and animation simulation is adopted to display video images and equipment status parameters of the fully mechanized mining face, which solves the problem that the shearer cutting image is not prominent.In terms of algorithm,a video compression method based on deep learning technology is proposed, in addition to compressing the video data itself, the inter-frame data is also compressed, which effectively reduces bit rate of the algorithm; the nonlinear anti-distortion model(NAM) correction algorithm is used to eliminate image distortion, the speeded-up robust features(SURF) detection algorithm is used for feature point detection, and image fusion is realized through bilinear interpolation method, so as to achieve panoramic video stitching. Development directions of video monitoring technology for fully mechanized mining face are discussed including camera self-cleaning technology, intelligent recognition technology, panoramic video stitching technology of working face, 5G and WiFi6 fusion communication technology, coal and rock interface identification technology.