A COMPARISION OF VARIOUS EDGE DETECTION TECHNIQUES IN MOTION PICTURE FOR IDENTIFYING A SHARK FISH

Abstract

The significant feature of detecting the motion image objects in this study it try identify the shark fish videos by removing the Background of the image. The main method involved in the detecting from the background is the foreground detection of the image. There are many techniques which usually ignore the fact that the background images consist of different image objects whose conditions may mostly change occur. In this study, a motion picture identification procedure is proposed for real time motion video frames by comparing the three key classes of methods for motion detection primarily the Background Removal (Subtraction) followed by the Temporal distinguishing (differencing) and Optical Flow method. Structured hierarchical background procedure is proposed based on segmenting background images objects. It mainly divided the background images divided into several parts (regions) by the Support Vector Machine (SVM) followed by a structured hierarchical model is built with the region procedure and pixel model procedure. In the region model method, the image object is extracted from the histogram of specific parts which is same to the kind of a Gaussian-combination model. In the pixel model procedure, it is been demonstrated by histograms, picture, which shows gradients sample of pixels in each parts based on the concurrent occurrence of object variations. In this study, it suggests Silhouette detection procedure and it is used. The experimental result are counter validated with a video database to illustrate its efficiencies, which is involved, from static to dynamic scenes by analyzing it with some distinguished motion detection methods chiefly Temporal differencing method followed by Optical Flow method and based on the outputs a motion detection procedure for real time video frames can be created which is cost effective, it shows good rate of accuracy, which is less rate of reliability in simple, less of complexity and well adapted to several kinds of shadow (image) distribution.