The Research Of 3-D Image Threshold Segmentation Algorithm

Image segmentation is a very important link in digital image processing.As an important basis for image recognition,analysis and research,the segmentation results will directly affect the subsequent process.In the existing segmentation methods,threshold segmentation is widely used because of its simplicity,high efficiency and easy for scholars to understand.Therefore,determining the optimal threshold is the key to threshold segmentation.In this aspect,domestic and foreign scholars have carried on the extensive research,and proposed the massive threshold value segmentation method.Otsu method has become the most widely used threshold segmentation method due to its wide application range,good segmentation effect and simple implementation.Because the existing image threshold segmentation algorithms have a variety of shortcomings,I used a new image threshold segmentation algorithm--Optimal Evolution Algo(OEA)for image threshold segmentation,and then used this algorithm for the research and experiment of 3d image threshold segmentation.According to the existing theories of biological evolution and genetic algorithm,the population evolution model and threshold updating model are improved.As the one-dimensional Otsu method only considers the pixel gray information and ignores the spatial correlation between pixels,it has some deficiencies,such as poor noise resistance.In the original one-dimensional method,the two-dimensional maximum entropy method considered the information of the mean value of the neighborhood,which improved the image segmentation results.However,because this method assumed the probability of the occurrence of the useful region as 1 in the threshold calculation process,this assumption resulted in serious loss of image information and inaccurate image detail segmentation.The 2-d Otsu method adopts post-processing scheme to improve this deficiency,making the segmentation performance strong.However,in the process of correcting the false segmentation,some important details of the image are lost.The method in this paper is based on the 3d histogram to determine the optimal threshold,and the algorithm has a strong segmentation performance.Because there is no computational complexity,the details of the image are relatively rich.The experimental results show that the segmentation performance of the proposed method is strong,and the segmentation results can preserve the important details of the image well.Therefore,the superiority and feasibility of the proposed method are verified from different perspectives.