Research On Infrared Image Segmentation Based On Improved Threshold

Aircraft skin is an important part of aircraft fuselage.Damaged skin will increase the safety risks of aircraft flight.Infrared NDT has the characteristics of high accuracy and easy acquisition in practical application.However,infrared images have poor resolution,low contrast and blurred vision.Traditional threshold segmentation algorithms often fail to achieve ideal segmentation results.In this paper,a random walk segmentation method based on the cuckoo search algorithm is proposed to segment the aircraft skin damage image.The random walk algorithm not only enjoys good segmentation effect even with interference or low signal-to-noise ratio,but also the advantages of fast segmentation,no iteration,and easy extension to multi-dimensional space.However,the random walk algorithm is sensitive to the number and position of the initial point.To solve the problem of point selection,a random walk segmentation method based on cuckoo search algorithm is proposed.Before the random walk segmentation,the number and position of the initial point should have been determined.In order to improve the computing efficiency of cuckoo search algorithm,inertia weight parameters are proposed to accelerate algorithm iteration.Experiments show that this method can effectively improve the efficiency of the algorithm.A two-dimensional Tsallis entropy multi-threshold segmentation method based on improved DE(Differential Evolution)algorithm is proposed to solve the problem of edge insufficient segmentation in the segmentation results of random walk algorithm.Tsallis entropy is a non-extensive entropy based on traditional Shannon entropy.Because there are many non-additive information in the image,the traditional segmentation algorithm is prone to appear insufficient segmentation.Aiming at the complex computation problem of the multi-threshold of two-dimensional Tsallis entropy,the improved DE algorithm is used to accelerate the optimization process.Experiments show that this method can not only effectively improve the problem of edge insufficient segmentation,but also speed up the computation efficiency of the algorithm.