Multi-Hypothesis Approach for Efficient Human Detection in Complex Environment

Date of Award


Degree Name

Ph.D. in Engineering


Department of Electrical and Computer Engineering


Advisor: Vijayan Asari


Over the last decade, detection of human beings become one of the most significant tasks in computer vision due to its extended applications that include human computer interaction, visual surveillance, person identification, event detection, gender classification, robotics, automatic navigation, and safety systems. However, this task is rather challenging because of the fluctuating appearance of the human body as well as the cluttered scenes, pose, occlusion, and illumination variations. For such a difficult task, most of the time no single-feature algorithm is rich enough to capture all the relevant information available in the image. To improve the detection accuracy, we propose a multi hypothesis approach containing various aspects of human visual perception. We explore the effectiveness of spatial domain behavior, phase domain behavior, and neighborhood dependency of an image for describing the object region. These cues will lead to the description of the shape and texture of specific objects. Shape features are extracted based on both the gradient concept and the phase congruency in LUV color space. The Center-Symmetric Local Binary Pattern (CSLBP) approach is used to capture the texture information of the image. The fusing of this complementary information yields to capture a broad range of the human appearance details that improve detection accuracy. The proposed features are formed by computing the phase congruency of the three-color channels in addition to the gradient magnitude and CSLBP value for each pixel in the image with respect to its neighborhood. Only the maximum phase congruency values are selected from the corresponding color channels. The histogram of oriented phase and gradients, as well as the histogram of CSLBP values for the local regions of the image, are determined. These histograms are concatenated to construct the proposed descriptor, which fuses the shape and texture features, and it is called the Chromatic domain Phase features with Gradient and Texture (CPGT). Human detection system based on the proposed descriptor (CPGT) is robust against illumination changes and is able to depict and detect the human objects in various scales, viewpoints, and postures as well as detection under partial occlusion and realistic environments. Several experiments were conducted to test and evaluate the detection performance of the proposed descriptor. The challenging and the well-known INRIA, DaimlerChrysler, and NICTA datasets are used in these experiments. A Support Vector Machine (SVM) classifier is used in these experiments to classify the CPGT features. The results show that the proposed algorithm has better detection performance in comparison with the state of art feature extraction methodologies.


Electrical Engineering, Artificial Intelligence, Phase congruency, human detection, color space, shape features

Rights Statement

Copyright 2018, author