Extraction of Visual Features from Video Sequences for Better Visual Analysis

Abstract

Video have a basic and non basic features, where basic features includes e. g. color, shape, size and non basic features include orientation of a image. Whereas Video Sequences is a series of shots/frames on a subject that are edited together to tell a story. Visual features describes the details about the image contents, which are used in various applications like, visual search, object recognition, image registration and object tracking. Many visual analysis task requires the features to be transmitted, thus it calls for the different coding algorithms to attain a target level of efficiency. Here an effort has been taken to implement a coding algorithm for local features extraction such as SIFT (Scale Invariant Feature Transform). The first stage comprises of using the SIFT algorithm property to find the ‘point of interest’ of an image. Further the use Kalman Filter algorithm is done as an application purpose of motion based single or multiple object detection and tracking.