Human Identification On the basis of Gaits Using Time Efficient Feature Extraction and Temporal Median Background Subtraction

Abstract

-Gait analysis is basically referred to study of human locomotion. From the surveillance point of view behavioral biometrics and recognition at a distance are becoming more popular in researchers rather than interactive and Physiological biometrics. In this paper, a time efficient Human gait identification system is proposed. Initially Human silhouettes are extracted by using temporal median background subtraction on video frames, which successfully removes shadows and models even complex background, proposed gait algorithm extracts contours from foreground silhouettes images and then three bounding boxes are drawn around contoured human image 1) upper part for arms movement 2) middle part for thigh and knee angles 3) Lower part for legs movement, knee and ankle angles. Gait cycles are extracted to find gait period and to take final decision for gait features selection, which is used for training. Thigh, Knee, Ankle angles and bounding boxes' widths are used as gait signatures but middle portion of human contains less variations of width in gait cycle hence computing efficiency can be achieved by ignoring width factor of middle part. SVM based training and identification is performed on extracted gait features. The proposed system is assessed using publicly available gait datasets and some indoor experimental videos created for this research work. The results reveal that the proposed algorithm is able to achieve an outstanding recognition rate.

Motion Segmentation from Surveillance Video using modified Hotelling's T-Square Statistics

“…Elgammal et al [10] proposed non-parametric kernel density estimation technique for background modeling. Median based background subtraction in temporal domain is proposed in [28].…”

“…Nearly, most of the statistical background subtraction based approaches [2,21,28,30,32,34] model the background as normally distributed. However, in the realtime environment, this assumption cannot be considered [10].…”

“…Several techniques are proposed in the literature for motion segmentation and these techniques can be categorized as conventional background subtraction [14], statistical background subtraction [2,10,21,28,30,32,34,37], temporal differencing [1,5,6], optical flow [9] and hybrid approaches [3,4,7,8,15,16,17,31,33,40].…”

Development of Algorithm to Reduce Shadow on Digital Image

“…This greatly affects geometrical properties and appearance of the object to be detected [3]. Shadow can cause inaccuracy in some applications, such as the applications for plant leaves segmentation [4], license plate recognition [5], gait recognition [6,7], analysis of remote-sensing images [8,9], classification of objects from video surveillance system [10], underwater object detection [11] and clustering [12].…”