An electronic image stabilization algorithm of cameras for, underground moving carriers

Abstract：The quality of a video is greatly influenced by the irregular vibrations from the cam- era for underground moving carriers. In this work, an electronic image stabilization algorithm is proposed to improve the stability and fidelity of video sequence from cameras for moving car- riers by removing the inter-frame vibration between video sequences. Firstly, the equations set to solve the vectors of motion are established with the moving transformation model. By ex- tracting the Harris corners, the mapping relationship between the candidate frame and the cur- rent frame is built based on the calculated results of the equations set as global motion vectors. Then, according to the uncertain vibrating positions of moving carriers, the motion offset vec- tors are computed by establishing the state--forecasting equations for the particles. The obser- vation value and its probability density function are defined by the 2D Euclidean distance. The experiments of validity and fidelity about algorithm show that the proposed algorithm can sepa- rate the active scanning motion vectors form random vibrations of camera, which achieves to improve the quality of output video sequence. The average peak signal to noise rate (PSNR) was increased to 25.38 DB after the vibrating video sequence was stabilized.