A Keyframe Extraction for Edited Video Summarization using Spatio -Temporal Analysis of GLCM based Texture Measures

Due to advancement of camera devices, huge amounts of digital video data are generated. To browse and analyze this video content, the user takes more time. In order to reduce the time to analyze and browse the video content, a summary is to be created. It can be achieved by extracting keyframes of the original video. This paper presents a summarization technique for edited video by using the GLCM representation. The unique content is determined for evaluating inter-frame difference by the following texture features such as Energy, Homogeneity, Mean, and Contrast. The pre-defined threshold values are set for each feature to select the appropriate keyframes. The summarization model is designed to capture the important properties of each frame. Specifically, this model is applicable to edited videos. This model was evaluated by two benchmark datasetsviz.,TVsum50 and SumMe. Finally, experimental results were compared with the benchmark existing algorithms to evaluate the efficiency of the proposed work.