Optimal Image Thresholding Via Intelligent Genetic Algorithm Based On Phase Correlation

Image segmentation is the key step of the process from image processing to image analysis, and is the basis of image analysis, recognition and understanding. Image segmentation, partitioning an image into different regions with some specific properties, has always been an important and challenging problem for many years. The application of image segmentation is very extensive, especially in the information society nowadays, using computer technology to carry out the automation process of the vision information processing makes it almost appeared at all relevant fields of image processing, and that involves various image types. It can be used in the realms of the image encoding, the processing and analysis of medicinal image, the analysis of transportation image, the automatic industry, the on-line product examination, the production line control, the documental image processing, the remote sensing and biomedical science image processing, the public security surveillance, the military affairs, the physical education and even in the agriculture engineering. Usually, image segmentation is the basis of image analysis, image recognition, image compression and image encoding; the accuracy of image segmentation's result will affect the work of the subsequence directly.Thousands of methods have been put forward. By now, there is no one method that is fit for all types of images.Genetic algorithm(GA) is a sort of efficient, paralled,full search method with its inherent virtues of robustness, parallel and self-adaptive characters. It is suitable for searching the optimization result in the large search space. Now it has been applied widely and perfectly in many study fields and engineering areas. In computer vision field GA is increasingly attached more importance. It provides the image segmentation a new and effective method.Algorithms and analyses about image segmentation are presented; An overview on the theories and the recent development is given. Also the status of GA applied in the image segmentation field is presented, and the theories, steps, results and analyses of several GA applied in the image segmentation are given; The theory of Fourier transform and its application in image processing fields also are discussed and analyzed in this paper.Nowadays, the popular image segmentation methods almost take part in the time space of the image, but in this paper, the processed images are transformed form time space to frequency space. Thus avoid dealing with isolated pixels in the image, and use correlated phase function as the judgment rule to search for the optimal threshold value from the global solution space. Phase-based optimal image thresholding can deal with poor contrast, high noise to signal ratio, complex patterns, and variable modalities in the gray-scale histograms, and it can obtain very good result. However, when the size of image increases, the speed of this process drops obviously. In order to overcome this shortage, a new intelligent genetic algorithm (IGA), which applies an intelligent crossover(IC) based on orthogonal arrays (OAS), is proposed to search the optimal combination of the thresholding that is finally used to segment the image. The results of experiments show that the proposed method can segment the image effectively and properly, as well as is robust to noise. Moreover, the proposed method reduces the computation time greatly.