The Research Of Image Interpolation Based On Non-Local Patches Collaging

With the rapid development of information society,high-quality images are in demand in the field of military remote sensing,astronomical observation,medical imaging,and public applications.As the image acquisition equipment was limited by the photoelectric sensor resolution and noise interference,some of the image high-frequency information may loss,reducing the quality of the image.Although professional hardware devices can get high-quality images,but the cost is substantial.Image interpolation technology has become an effective way to solve this problem.Since natural image has information redundancy,we can use this feature to carry on the high resolution interpolation reconstruction to the low resolution image.Although the traditional linear interpolation algorithm can improve the quality of the image,but it has obvious edge blur and jagged phenomenon.Although newly proposed interpolation algorithm can further improve the interpolation effect of the image,the higher computational complexity limits their application range.In this thesis,the non-local Patches Collaging Interpolation(NPCI)algorithm is updated,improving the effect of image high resolution reconstruction.The algorithm is not only simple in principle,but also has low computational complexity.The main work of the thesis is as follows:(1)Introduce the background and significance of image interpolation algorithm research,combined with the relevant research progress at home and abroad,the current mainstream interpolation method of the principles and characteristics of the summary.(2)By introducing the weighting matrix,the original two different restoration equations in the NPCI algorithm have a unified expression.Since the weighting matrix can adjust the weight ratio of the sampling grid point and the non-sampling grid point,this changes the weight of the sampling grid point so that the interpolation effect can be improved.(3)The thesis proposes to use the absolute difference(measure)between image blocks to measure the similarity between image blocks.Compared with the inner product measure in the original algorithm,this measure reduces the computational complexity of the NPCI algorithm.(4)The value of parameter K in the NPCI algorithm is selected empirically,which does not take into account the fact that the blocks to be constructed will require different numbers of local image blocks to be reconstructed due to differences in texture features.The thesis uses the dynamic strategy to determine the K value to enhance the adaptability of the algorithm.The experimental results show that the improved NPCI algorithm has a better effect than the original algorithm.By comparing the interpolation effect of other algorithms,it can be found that the improved algorithm has better effect on the simulated interpolation of each standard test image.