Image Inpainting Based Onneighbour Information

This paper proposed a model based on pruning samples by referring to four-neighbors,for enhancing the efficiency and accuracy of pruning samples.Meanwhile,in order to clearly distinguish the main structure edges and abundant texture region,image inpainting algorithm based on neighboring window feature value is proposed.Expatiation of improvement is following:1)Image inpainting based on pruning samples referring to four-neighbors was proposed,firstly,in order to solve the ignores that Sum of Squares of Deviations of pixel method for structure edge information,edge matching constraint was introduced into matching method,which improves samples matching precision.Then,priority formulas was obtained by introducing structure weight,structure intensity factor,confuse value and confidence.Finally,the efficient neighbors was chose and referred to pruning samples by computing overlapping information between target block and neighborhood blocks patches.The experimental results demonstrate the Peak Signal-to-Noise Ratio(PSNR)of the improved algorithm is increased by 0.5 dB to 1 dB compared with the contrast methods with speeding up inpainting process,the recovered image is much continuous for human vision.2)Image inpainting algorithm based on feature value of neighboring window was proposed,firstly,taking advantage of mean gradient matrix to extract feature points,and computing the gradient direction feature and gradient richness for neighboring window feature value by analyzing their gradient features.Then,improved priority formula which can be distinguished the main structure edges and abundant texture region effectively was got.It made inpainting the sequence and consequence of image inpainting more reasonable.Meanwhile,patch matching method was improved by combining proposed method and introduced structure different operator and enhanced matching accuracy.The experimental results demonstrate that the main structure edges and abundant texture region can be distinguished by neighboring window feature value,inpainting order was more reasonable,and the proposed method can overcome problems like texture blurring and structure dislocations and so on,the Peak Signalto-Noise Ratio(PSNR)of the improved algorithm is increased by 1 d B to 2 dB compared with the contrast methods with speeding up inpainting process,the recovered image is much continuous for human vision.