Movement Estimation in a Key Frame for an Effective Video Retrieval System
Abstract
Movement estimation in several frames introduces tremendous computational intricacy. This paper presents a fast full search Movement estimation algorithm for several reference frame Movement estimation. The proposed method can be applied to successive elimination algorithm (SEA) or its modified version, multi level successive elimination algorithm (MSEA). We have derived an extra inequality to eliminate the highly unfeasible person blocks in reference frames that are temporally preceding the initial reference frame. The experimental results show that the proposed method reduces computational intricacy while maintaining same quality as the full search algorithm.