Research On Infrared Image Segmentation Algorithm Based On Level Set

Infrared imaging system can work all day and the imaging advantages under extreme conditions make it more and more widely used in military and civilian fields.As the basis of target detection,recognition and tracking technologies,infrared image target segmentation can extract useful information from the image.Infrared imaging scenes are complex and diverse,the target gray distribution is uneven,and there are weak edges,which increase the difficulty of infrared image segmentation.The level set algorithm,which relies on mature and complete mathematical theories,is easy to solve and can achieve multi-objective topological segmentation,which makes it widely used in the field of image segmentation.However,the level set algorithm has a strong dependence on the initial contour.The initial contour far away from the target of interest will reduce the efficiency of image segmentation.At the same time,the effect of segmenting infrared images with uneven gray distribution and weak edges is not ideal.In response to the above problems,this paper improve the fuzzy clustering and boundary indicator function in the level set algorithm based on fuzzy clustering.The main content and innovations of this paper are as follows:(1)In order to set the initial contour of the level set algorithm in the target area more accurately,the fuzzy clustering algorithm is improved and the initial contour of the level set is set according to its segmentation result.Determine the number of cluster centers and cluster numbers of the fuzzy clustering algorithm according to the characteristics of the gray distribution of infrared pedestrian images,and automatically set the level set algorithm according to the cluster membership matrix corresponding to the largest cluster center of the clustering result Initial outline.Experiments show that the initial contours set in this way are located inside the real target,which is a suitable initial contour for the level set algorithm.(2)In order to improve the segmentation accuracy of the level set algorithm for uneven gray distribution images,an adaptive weight coefficient is constructed according to the difference between the local median value of the pixel and the gray value of the cluster center to adjust the global and local based The proportion of the level set image segmentation.Comparative experiment results show that the improved algorithm has higher segmentation accuracy for infrared images with uneven gray-scale distribution in different backgrounds.(3)In order to further improve the segmentation accuracy of the infrared image with fuzzy weak edges by the level set algorithm,the gray value of the infrared image target and background is adjusted,and the concept of image entropy is introduced,and the boundary is changed by establishing an improved boundary indicator function.The information is integrated into the region-based level set image segmentation algorithm.Comparative experiments show that the improved algorithm has stronger ability to detect weak edges.From the perspective of subjective and objective quantitative analysis,it can be proved that the improved level set image segmentation algorithm in this paper can adaptively set the appropriate initial contour of the level set.Under different backgrounds,the infrared pedestrian image with uneven grayscale distribution and weak edges can be segmented with good segmentation effect.