Real-time image-based motion detection using color and structure
Motion detection is an important problem in computer vision and has multiple applications in the real world, including surveillance-related activities, gaming, and human-robot interactions. These systems need to be robust enough to handle fluctuations in light intensity and other external factors like noise and compression artifacts. In this thesis a method is proposed for detecting the regions of motion from a video sequence in real time. The main idea of this work is to detect motion based on both structure and color. Structure-based detection is carried out using information from the Census Transform computed on gradient images obtained with Sobel operators. The Census Transform characterizes local intensity patterns in an image. Color-based detection is done using color histograms, which allow efficient characterization without prior assumptions about color distribution in the scene. The probabilities obtained from the gradient-based Census Transform and Color Histograms are combined in a robust way to detect the active motion zones. Experimental results demonstrate the effectiveness of this approach. ^ This thesis also presents an application for motion detection in foveal visual systems. Foveation reduces power and bandwidth requirements in a system. But the main challenge is to identify the region of interest that must be transmitted in high-resolution format while maintaining the rest in low resolution. In this work this issue is addressed, by proposing motion detection as a cue to select the region of interest in a given scene. Preliminary results are also presented to show the successful application of this approach. ^
Chakraborty, Manali, "Real-time image-based motion detection using color and structure" (2009). ETD Collection for University of Texas, El Paso. AAI1473855.