Research On Target Tracking Algorithm Based On Multiple Methods

Researching on target tracking algorithm based on multiple methods,the main research present situation carries on the analysis from three aspects: Firstly,adaptive image enhancement based on multiple algorithm fusion.Secondly,study on mobile target real-time image registration based on improved SURF algorith.Thirdly,neurofilament protein serves as the marker of the state for ALS(Amyotrophic Lateral Sclerosis)in medical filed.Firstly,adaptive image enhancement based on multiple algorithm fusion.For the detail features of the complex degraded images being effectively restored in engineering application,an image enhancement method of the multiple algorithm-fusion is introduced in this paper.This fusion algorithm is besed on the thory of digital image and integrates the advantages of the Laplace transform,Sobel gradient,box filtering,unsharp masking filter method and gray exponential law strength into the algorithm to enhance adaptively the fuzzy images.The original image,firstly,is decomposed into a base layer,a detail layer and a edge character layer by the Laplace filter and gradient filter;Secondly,the tiny details and edge characteristic information are enhanced and base information is compressed.Then the three layers of the image are processed smoothly and the noise is filtered by the box filtering.Finally,the dynamic range of gray level image is increased by the gray-scale transformation and the unsharp masking method,and the enhanced image is obtained.And under the same with load conditons,the new method is compared respectively with the traditional algorithms of the HE,AGCWD and WT.The experimental results that this method can effectively treatment the complex images with infection,and the sharpness of the image is increased by 13.1% ~ 126.1%,and avoid the phenomenon of excessive image enhancement,and obtain a superior subjective visual detail effects.Secondly,study on mobile target real-time image registration based onimproved SURF algorith.For the unstability of the real-time visual image characteristic point extraction and matching for the mobile target,an improved registration method of the multiple algorithm-fusion was introduced.Firstly,the method of bilateral filtering,Canny edge detection and morphological processing,was adopted to get the more robust image's edge map,and then the founded points were limitted in this edge map,then the Hessian matrix of SURF based on the discrete Gaussian-Hermite moment was modified and the character description vector was redefined in the algorithm.Following the above analysis,the Kendall coefficient constraint was discussed in image matching characteristic points.Secondly,the spectral radiant color invariant model and the I_SURF algorithm were used to match the real-time color image.Finally,the improved algorithm was combined with the update algorithm of adaptive target to match the mobile target in indoor environment.The experimental results show that the static image registration accuracy of the improved algorithm is higher than that of the SURF algorithm,and also the stability of the mobile image feature points extraction and the registration number have achieved over 97% under different rotating scales.Thirdly,neurofilament protein serves as the marker of the state for ALS(Amyotrophic Lateral Sclerosis)in medical filed.In order to accurately capture the activity characteristics of protein in the neurofilament,a new-type algorithm based on the particle filter of multiple methods-fusion is introduced.This fusion algorithm draws on color histogram,kernel function method and graph model strength to automatically trace neurofilament protein.In addition,for the sample impoverishment,namely,the situation that the majority of particles overlap on one single point in the computation of particle filter,resampling is utilitied,but anisotropies of some particles are prone to lose during this process,which may leads to low tracing precision,or even failure of trace,so a new binding method of resampling,on the basis of standard filter,is put forward.The results ofexperiment indicate,the algorithm based on the improved method of resampling and particle filter of multiple methods-fusion can effectively reduce the sample impoverishment,and,with high precision,track the moved and transmutative neurofilament protein,which presents a new supporting method for the research on neurofilament protein in medical filed.