Colony Image Segmentation Based On Genetic Algorithm

Processing of colony image is a basic and important job for quality inspection in agriculture, food and medical treatment analysis. For example, the number of colonies is an important quality parameter in evaluating the water pollution. In the past, this job is mainly operated in the form of handwork, and the process is complicated, time-consuming and low-efficiency. Furthermore, the result tends to subjective, error distinctive, and bad recurring. Image processing and analysis can free operator from the hard work, it also can improve the counting and analysis accuracy highly. All these fall to a result that the image processing and analysis is widely applied and then becomes one of the hottest research subjects in resent years.Image segmentation is an important step in colonies image processing and analysis. The segmentation result directly influences the result in the counting accuracy and the follow processing. But segmentation is a classic difficult problem in image information engineering, there is still no such an algorithm which can be applied to any image and get the best result. In colonies images, various in size and modality, and tend to cluster into each other, so there are more difficulties in segmentation. For this, this dissertation analyzed the presented technologies of segmentation, considering that the problem of image segmentation is treated as one of combinatorial optimization, genetic algorithm and simulated annealing algorithm can be used here, but it is seeing to suffer from several limitations when using them respectively. This dissertation proposed a combined algorithm which may be advantageous in combining the advantages of both GA and SA procedures while alleviating their individual shortcomings. By analyzing the colonies images which have various appearance and characteristics, using this technology of images segmentation to operate this experiment, researching a technology of automatic segmentation, this dissertation studies manage to promote the precision and speed of segmentation.This dissertation researches the technology of segmentation and the present condition of genetic algorithm, and tries to apply genetic-segmentation algorithm to the system of automatic reorganization. The function of this system is as follows: 1. Image filteringThere is middle and average filtering. The purpose of filtering is to promote the effect of segmentation and splitting, and the precision of counting.2. Image segmentationFirstly, converting color images to grey by combined algorithm which is based on genetic and simulated annealing algorithm. Secondly, filling the holes by filling algorithms, there are several algorithms such as seed-filling and polygon-filling. Finally, excluding the noise and splitting the over-lapped colonies.3. CountingThere are two methods. For grey images, it will have good result after thresholding, seed-filling, erosion and thinning, then it can carry on counting directly.