Fractional-Order Image Segmentation for Security Surveillance

The enhancement of image processing techniques related to security surveillance issues is considered a pressing demand nowadays. Everything is now documented by digital images, out of which important information is extracted. In this work, fractional-order edge detection filters are employed in edge-based Active Contour segmentation technique for noisy surveillance images. The fractional-order filters add extra degree of freedom, allowing more details to be detected in images, and enhancing the quality of segmented noisy images. Two types of noise, Salt and Pepper noise as well as Gaussian noise, are applied to test the noise performance of the presented segmentation technique. The superiority of the fractional-based segmentation over the conventional integer-based one was proven visually and numerically using peak signal to noise ratio for both types of noise.