A RATE-DISTORTION BASED APPROACH TO SHOT ANALYSIS AND REPRESENTATIVE FRAME EXTRACTION FOR DIGITAL VIDEO

• Main Authors: Min-Shu Chen, 陳民樹
• Other Authors: Jin-Jang Leou
• Format: Others
• Language: en_US
• Published: 1999
• Online Access: http://ndltd.ncl.edu.tw/handle/19717106056135949572

碩士 === 國立中正大學 === 資訊工程研究所 === 87 === In this study, a rate-distortion based approach to shot analysis and representative frame (r-frame) extraction for digital video is proposed. The objective is to partition a video sequence into a predetermined number L of shots and extract the corresponding L representative frames from the L shots so that the video sequence can be described compactly by the L representative frames and some auxiliary information. In the proposed approach, a video sequence is roughly partitioned into a number L' of shots, which is modified to a predetermined number L of shots. After the input video sequence is roughly partitioned into L shots, the proposed rate-distortion based approach to optimal shot boundary determination is used to determine the L-1 optimal shot boundaries (including camera breaks and gradual transitions) of the video sequence. The rate-distortion based approach to representative frame determination is used to extract the optimal L representative frames for the L shots of the video sequence. Based on the simulation results obtained in this study, the proposed approach can partition a video sequence into a predetermined number L of shots and determine the optimal L representative frames for the L shots. In terms of several performance measures, the performance of the proposed approach is superior to several existing approaches for comparison. This shows the feasibility of the proposed approach.