Video clip retrieval: An integrated approach based on WLD and SFTA

Abstract

Due to massive growth in digital media for communication, the number and size of videos being generated and accumulated is increasing and hence searching desired content from the video is becoming more complex. As text in videos provide additional information on the content of the videos, methods for text based retrieval is desired. This paper presents an approach to retrieve video clips having similar text from a large collection of videos. The text contained in the keyframe is localized using Weber Local Descriptor and texture feature obtained through Segmentation-based Fractal Texture Analysis from the localized text region is used for retrieval task. The approach is evaluated by conducting experiments on a set of videos. The results obtained are encouraging and indicate the effectiveness of the proposed approach.