Study On Image Segmentation Algorithm Based On Fuzzy Logic And Region Merging Algorithm

Digital image processing is a kind of technology which can finish the transformation from original picture to digital signal.It is a process to analyze and handle pictures by computer.The technology is widely used in many fields such as industrial production,medical image,security monitoring,image coding,satellite remote sensing and military,sports,agriculture and so on.Therefore,the researching of analyzing digital image is one of the focuses in the computer science while the image segmentation is also one of the most popular research directions in the field.It is a necessary condition for the computer to realize the technology of feature extraction and target recognition analyzing on pictures.The result of image segmentation will directly affect the subsequent analysis and processing of image.Firstly,the relevant theoretical knowledge is briefly introduced in this paper,including the application of morphology in image segmentation,the basic theory of watershed transformation,the regional merging algorithm based on gray similarity and the theoretical basis of fuzzy logic algorithm.Secondly,the morphological theory is mainly used in the image preprocessing,the image filtering and gradient transformation.The purpose is to reduce the noise to the next step,reduce the computational complexity.The watershed transform is used to initialize and segment the image.But due to the watershed transformation is easy to occur over-segmentation,it is necessary to continue to process the initial segmentation on the image.Regarding to the watershed transformation brings catchment basin and results to over-segmentation on tiny change,the paper states the methods which can obtain the closed area by it and restrain the issue by regional merging.In the first place,the regional adjacency graph is used to represent the topological relation between the regions in the image obtained by the watershed transformation,and the gray mean value of the region is calculated.Then the gray scale similarity between adjacent regions is taken as the weight of the arc of the adjacent graph.At last it can regionally merge through the gray-based similar regional merging algorithm on the image.In order to further improve the accuracy of the region merging algorithm,this paper brings to the concept of freeman code,which improves traditional the curvature of the edge evaluate function by using the feature of freeman code difference is proportional to the curvature of a point on the edge and reduce time and complexity of account the curvature of the edge evaluate the average rate of change about freeman code difference to describe the edge is regular or not.Meanwhile,in order to resolve overall evaluation of multi-index,this paper proposes an improved regional merging algorithm based on fuzzy logic.Through the reasoning method of fuzzy logic,we can collect the edge validity and grey-level similarity of two neighbouring regions.The consolidating degree can also be figured out according to the fuzzy rules and off-fuzziness output.During the process of regional merging,we can have a supervision and direction on it depending on its consolidating degree.The algorithm working on regions merging based on fuzzy logic reasoning is closely similar to the thinking process of human beings when inferencing and deducing the merging areas.The outcome of uniting regions not only ensures the presence of significant structure information,but also reduces the possibility of useless areas.It is a better way to achieve the aim of separating images that improves the generality of shapes of objects in segmentation results.