4. Case Study II: Image Segmentation

Image segmentation has been investigated as a vital task in a wide variety of applications including (but not limited to): document image analysis for extraction of printed characters; map processing in order to find lines, legends, and characters; topological feature extraction for extraction of geographical information; remote sensing image analysis; and quality inspection of materials where defective parts must be delineated among many other applications (Ghamisi et al. IEEE International Geoscience Remote Sensing Symposium (IGARSS) 2012). In addition, for the purpose of image classification and object detection, the use of an efficient segmentation technique plays a key role. This chapter is devoted to one of the important application of FODPSO, which is related to introducing a novel thresholding-based segmentation method based on FODPSO for determining the n − 1 optimal n-level threshold on a given image. This approach has been widely used in the literature for the segmentation of benchmark images, remote sensing data, and medical images. This chapter first, elaborates the mathematical formulation of thresholding-based image segmentation. Then, some well-known thresholding segmentation techniques such as genetic algorithm (GA)-, bacteria foraging (BF)-, PSO-, DPSO-, and FODPSO-based thresholding-based segmentation techniques are compared in terms of accuracy and CPU processing time. Experimental results demonstrate the efficiency of the FODPSO-based segmentation method compared to other optimization-based segmentation methods when considering a number of different measures.