Human Object Classification Using Dual Tree Complex Wavelet Transform and Zernike Moment - Transactions on Large-Scale Data-and Knowledge-Centered Systems - page 87

Human Object Classification Using Dual Tree Complex
Wavelet Transform and Zernike Moment
Manish Khare 1 , Nguyen Thanh Binh 2(  ) , and Rajneesh Kumar Srivastava 1
1 Department of Electronics and Communication, University of Allahabad, Allahabad, India
{mkharejk,rkumarsau}@gmail.com
2 Faculty of Computer Science and Engineering, Ho Chi Minh City
University of Technology, Ho Chi Minh, Vietnam
[e-mail]
Abstract. Presence of variety of objects degrade the performance of video sur-
veillance system as a certain type of objects can be misclassified as some other
types of object. Recent researches in video surveillance are focused on accurate
classification of human objects. Classification of human objects is a crucial
problem, as accurate human object classification is a desirable task for better
performance of video surveillance system. In this paper we have proposed a
method for human object classification, which classify the objects present in a
scene into two classes: human and non-human. The proposed method uses
combination of Dual tree complex wavelet transform and Zernike moment as
feature of object. We have used support vector machine (SVM) as a classifier
for classification of objects. The proposed method has been tested on standard
dataset like INRIA person dataset. Quantitative experimental results shows that
the proposed method is better than other state-of-the-art methods and gives bet-
ter performance for human object classification.
Keywords: Object classification · Zernike moment · Dual tree complex wavelet
transform · Support vector machine
1 Introduction
Human object classification in real scenes is a challenging problem with many useful
applications like object tracking, object identification, etc. [1-2]. Goal of any object
classification algorithm is to develop a method having capability to interpret the objects
into different groups. Object classification algorithm must work under real time con-
straints and robust in the situations like, color variation in human cloths, variation in
natural conditions, different size of human objects, etc. Feature selection and machine
learning techniques are essential components of any classification algorithm [3]. Object
classification algorithm is commonly divided into three components: (i) selection of
feature, (ii) extraction of feature, and (iii) classification. Correctness of any classifica-
tion scheme lies on the selected feature, therefore selection of effective feature is a cru-
cial step for successful classification. Machine learning based methods have been used
for developing object classification algorithm. The learning in a classification system is