ROI-centered compression by adaptive quantization for sports video

Abstract

In this paper, we propose a region of interest (ROI)-centered compression technique by adaptive quantization for sports videos. The loss of visual detail in an ROI in a broadcast soccer video is solved as seen in DMB (Digital Multimedia Broadcasting) through the mobile device. For tracking an ROI, in a soccer video, we track ball and player candidates using the Dynamic Kalman Filter (DKF) and send information of ROI macro-blocks (MBs) determined by tracking results to an encoder. The encoder allocates quantization step size (Qp) to ROI and non-ROI MBs based on priority and controls bit-rate using proposed rate-control (RC) algorithm. Experiment results show the proposed method increased visual detail in the ROI while maintaining a similar bit rate with the uniform quantization method.