Research On Improvement And Application Of Otsu Image Segmentation Method

As the increasingly development of intelligence and things technology,machine vision is playing an dramatically significant role in every field,which puts forward higher and higher requirements for image processing technology.As an intermediate step in the process of image processing,image segmentation has an important impact on the results of image processing and recognition.Although there are many kinds of image segmentation methods,threshold method is the most classical one,among which Otsu method is a representative one.The principle of Otsu method is simple and easy to implement,which has been widely used in the field of image processing.However,with the processing of increasingly complex images,Otsu method has exposed a variety of shortcomings.Therefore,this paper focuses on the Otsu method,a classical segmentation algorithm.In this paper,firstly,we deeply analyzed the essence of Otsu method by using set mapping principle,and explored the composition,design and partition methods of histogram.Secondly,we discussed the mechanism of noise detection and correction,and studied the algorithm's complexity and the fast segmentation methods.Finally,the application of the proposed method is tested by using gesture image and brain MR image.The specific research contents of this paper are as follows:1.Although line intercept histogram Otsu method has good segmentation performance,its computational efficiency and anti-noise capability still need to be improved.So this paper analyzed the set mapping essence in line intercept histogram Otsu method,and then proposed a new Otsu method based on set mapping and trapezoid region intercept histogram.Firstly,establishing mapping rules to map the two-dimensional(2D)histogram pixels to different trapezoid regions,and compressing sufficiently the threshold space to improve the algorithm efficiency.Then,introducing the post-processing strategy to improve the mapping rules so that the algorithm can give consideration to both anti-noise capability and detail retention.Experimental results show that the algorithm has good performance in computational efficiency,and can achieve a balance between anti-noise capability and detail retention in image segmentation.2.The traditional 3D Otsu algorithm has high computational complexity,and the anti-noise capability needs to be improved.In this paper,by using a plane perpendicular to the main diagonal to partition the 3D histogram,designing a plane intercept histogram and giving the corresponding between-class variance criterion,thus proposing an Otsu method based on plane intercept histogram and geometric analysis.In order to enhance the algorithm's anti-noise capability,based on the geometric analysis principle,the method reclassified the noise in regions 2-7,which was ignored by the traditional 3D Otsu method.The test results show that compared with the traditional 3D Otsu method and several improved methods,this method has obvious advantages in algorithm efficiency and segmentation effect.3.The threshold value of Q in the post-processing of traditional cross section projection Otsu method is a preset constant,which is not suitable for images with different noises.To solve this problem,this paper proposed a multi-objective cross section projection Otsu method based on memory knetic-molecular theory optimization algorithm.Based on the maximum between-class variance criterion and the maximum peak signal to noise ratio(PSNR)criterion,a multi-objective image segmentation model is established to take into account the segmentation accuracy and anti-noise capability for image segmentation by combining threshold Q with segmentation threshold T.In order to improve the efficiency of the algorithm,a memory knetic-molecular theory optimization algorithm is proposed for the multi-objective cross section projection Otsu method by introducing the artificial memory principles into knetic-molecular theory optimization algorithm.The experimental results show that this method has significant advantages in segmentation accuracy,anti-noise capability and robustness,and is more universal applicability for images with different noises.4.By analyzing the essence and deficiency of the improved Otsu method,this paper proposes a noise adaptive angle threshold based Otsu method for gesture image segmentation.It first designs a 2D histogram of gray value-neighborhood truncated gray mean to avoid the interference of extreme noise by discarding the extremes of the neighborhood.Then,the probability that the pixel is noise is calculated according to the actual situation,adaptive filtering is implemented to enhance the algorithm's universal applicability.It finally converts the threshold space to an angle space from 0° to 89°,and the threshold search range is compressed to improve its efficiency.As the gesture is close to the background and the boundary is blurred,this paper combines the global and local Otsu method to segment the gesture images based on the angle space.On the one hand,it uses the global Otsu method to obtain the global threshold t1.On the other hand,it uses the local Otsu method to obtain the local threshold t2,and segments gesture images based on t2.Experimental results show that the proposed method is effective and can accurately segment gesture images with different noises.5.Otsu method is a common method in brain MR image segmentation,however,in image segmentation,the current Otsu method is often difficult to take both accuracy and anti-noise capability into consideration.In this paper,by improving the trapezoid region intercept histogram Otsu method,an adaptive trapezoid region intercept histogram Otsu method is proposed.On the basis of bilateral filtering,the method uses Sigmoid function to identify the noise and adaptively calculate the weight of neighborhood pixel,and then constructs the 2D histogram of gray scale-adaptive weight neighborhood gray mean to enhance the algorithm's anti-noise capability and detail retention.The double threshold model is adopted,firstly,the macro-threshold T1 is solved based on the trapezoid region intercept histogram Otsu method.Then,the micro-threshold T2 is determined by the inter-class variance criterion again in the trapezoid region corresponding to T1.The image is segmented according to the value of T2 to improve the accuracy of image segmentation.Based on neighborhood information,designing an adaptive parameter l to identify and correct noise,thus enhancing the universality of the algorithm.The experimental results show that the proposed method is effective and can be well applied to MR image segmentation.Otsu method is a classical image segmentation method,at present,it is still widely studied and applied.Based on this topic,this paper carries out thorough research,analyzes the principle,essence and shortcomings of Otsu method,and proposes five improved Otsu methods,which show good segmentation performance for engineering practice images such as gesture images and brain MR images.It can be seen that the related work in this paper can not only enrich the theory of image segmentation algorithm,but also have a good prospect of engineering application.