An implementation of the video retrieval system by video segmentation

Image information is widely used for the content-based retrieval of the image sequence. It is mainly used to segment a video by scene. Through this task, the structural video browsing can be achieved. The process that divides video into shots is called ldquovideo segmentationrdquo. For the video segmentation, detecting cut which is turn point of scene is called ldquocut detectionrdquo. In this paper, for the segmenting video, we use two MPEG-7 visual descriptors, which are the HMMD (Hue-Max-Min-Diff) color model and the EHD (Edge Histogram Descriptor). The goal of this paper is to implement the retrieval system as hardware by Verilog-HDL. We perform the FPGA verification, and implement the proposed retrieval system for the image sequence.