Research On The Model For Segmenting Groups Of Similar Images Based On Partial Differential Equation

The ACGS(Active Contours With Group Similarity)model which is based on the C-V model and combines with the matrix's low-rank constraint can play a good role in segmenting the groups of similar images in which the features of the object is missing or misleading,but it performs poorly on the groups of similar images with intensity inhomogeneity.In this paper,we mainly analyze and research the existing problems of the ACGS model and then propose the corresponding improved models to get better segmenting results.For one thing,in order to overcome the problem that the ACGS model has a bad result on the groups of similar images with intensity inhomogeneity,the active contour model with dual contour evolutional curve that is based on the LBF model is introduced to replace the C-V model in the ACGS model.For another,although the improved model above segments the inhomogeneous images with slowly changing intensity well,it performs badly in the inhomogeneous images with dramatically changing intensity.To solve this problem,we combine the model with local InH consistency energy of the lnH_ACM model by setting weight parameters,and thus introduce the pixel inhomogeneity factor(PIF)to estimate whether a pixel belongs to the uneven area with dramatically changing intensity in the image.The specific jobs in this paper are as follows:(1)The model for segmenting the group of similar images based on dual contour evolutional curve is proposed.To overcome the bad result of the groups of similar images with intensity inhomogeneity segmented by the ACGS model,we introduce the active contour model with dual contour evolutional curve based on the LBF model to replace the C-V model in the ACGS model.The improved model firstly keeps the low-rank constraint condition to maintain the similarities between the images,secondly combines the LBF model to segment the images with intensity inhomogeneity,and finally overcomes the problem that the LBF model is sensitive to the initial contour by the interaction between the two curves during the evolution.From the experiment results,we can see that compared with the C-V model?LBF model?ACGS model and the active contour model based on dual contour evolutional curve,the proposed model can get a better result for segmenting the groups of similar images with intensity inhomogeneity.(2)The dual contour curve model combined with low-rank constraint and pixel inhomogeneity factor is proposed.In order to solve the problem that the model for segmenting the group of similar images based on dual contour evolutional curve has a bad performance in segmenting the inhomogeneous images with dramatically changing intensity,we combine the model with local InH consistency energy of the lnH_ACM model by setting weight parameters,and thus introduce the pixel inhomogeneity factor(PIF)that is benefit to segment the images with texture intensity to make the model's result better.The experiment results show that,compared with the model for segmenting the group of similar images based on dual contour evolutional curve,the lnH_ACM model and the lnH_ACGS model,the new model has the best result in segmenting the images both containing the regions with dramatically changing intensity,and the regions with slowly changing intensity.