Background subtraction is the technique of segmenting moving foreground objects from stationary or dynamic background scenes. Background subtraction is a critical step in many computer vision applications including video surveillance, tracking, gesture recognition etc. This thesis addresses the challenges associated with the background subtraction systems due to the sudden illumination changes happening in an indoor environment. Most of the existing techniques adapt to gradual illumination changes, but fail to cope with the sudden illumination changes. Here, we introduce a Global change reactive background subtraction to model these changes as a regression function of spatial image coordinates. The regression model is learned from highly probable background regions and the background model is compensated for the illumination changes by the model parameters estimated. Experiments were performed in the indoor environment to show the effectiveness of our approach in modeling the sudden illumination changes by a higher order regression polynomial. The results of non-linear SVM regression were also presented to show the robustness of our regression model.

Integration of audio and video signals for automatic speech recognition has become an important field of study. The Audio-Visual Speech Recognition (AVSR) system is known to have accuracy higher than audio-only or visual-only system. The research focused on the visual front end and has been centered around lip segmentation. Experiments performed for lip feature extraction were mainly done in constrained environment with controlled background noise. In this thesis we focus our attention to a database collected in the environment of a moving car which hampered the quality of the imagery. We first introduce the concept of illumination compensation, where we try to reduce the dependency of light from over- or under-exposed images. As a precursor to lip segmentation, we focus on a robust face detection technique which reaches an accuracy of 95%. We have detailed and compared three different face detection techniques and found a successful way of concatenating them in order to increase the overall accuracy. One of the detection techniques used was the object detection algorithm proposed by Viola-Jones. We have experimented with different color spaces using the Viola-Jones algorithm and have reached interesting conclusions. Following face detection we implement a lip localization algorithm based on the vertical gradients of hybrid equations of color. Despite the challenging background and image quality, success rate of 88% was achieved for lip segmentation.