COLOUR, TEXTURE AND SHAPE FEATURE ANALYSIS FOR PERSON RE-IDENTIFICATION TECHNIQUE

ABSTRACT Person re-identification across different surveillance cameras has become one of the recent andchallenging approaches in the area of video surveillance system. Although many methods have beenintroduced for re-identification, certain issues like illumination, pose variation and occlusion are unsolved.All the existing methods focused only on developing learning methods by varying the features. The aim ofthis work is to analyze the effectiveness of feature for re-identification. The colour, texture and shapefeatures has taken for the analysis. The images are collected from multiple databases for training andtesting. The analysis shows that every feature has its unique properties and works according to that tosolve the issues KEYWORDS Re-identification, Colour, Texture, Shape ,Surveillance 1. INTRODUCTION A fundamental task of a distributed multi-camera surveillance system is to identify the peopleacross camera views at different locations and times. This is known as the person re-identification(re-id) problem and it is used in many crucial applications such as Criminal identification,videosurveillance, access control, ID Card and forensic search . The purpose of re-identification systemis to accurately recognize a person's identity from the cameras in a distributed network and takesome actions on the output of the recognition process. The classical biometric system is solvingthis problem using finger print, palm, iris and face [8]. But these features cannot be use forautomatic environment like surveillance network in public places. It is necessary to concentrateon the full appearance of persons.