Analysis of Daubechies Wavelet Transform Based Human Detection Approaches in Digital Videos

Human detection in digital videos is used in a variety of real time scenarios such as unusual activity detection in crowded places, gender identification and age determination and people detection in heavy traffic. Every human detection algorithm detects all the moving objects in the initial stage. The next step is to differentiate the moving object as a human being (or) non-human being. This classification is based on shape, texture and motion features. The objective of this work is to compare the performance of two different human detection algorithms. One is the human detection based on shape and another one is human detection based on Daubechies wavelet Transform. The shape based detection uses the shape information of human body to classify the moving objects. Daubechies wavelet transform is shift invariant in nature. Thus, the algorithm is able to detect even small hand or head movements. The performance of these two algorithms are compared in terms of detection accuracy, Precision and Recall. Experimental results have shown promising results for the wavelet transform based approach.