Image Interpolation Algorithm Based On Multi-Direction And EDGE-Preserving

With the rapidly development of multimedia technology and computer network, image interpolation technique was widely used in military, medical, aerospace and other fields. High-resolution images can provide accurate and detailed information, play a positive role on people’s work and life. Hence, this thesis was focus on the image interpolation technology that was applied to overcome the deficiency of the existing algorithms, and obtain high resolution image well.This work mainly include three parts of work. Firstly, this part analyzes the image interpolation technology both at home and abroad, and recommends the existing image interpolation algorithms. The classification of interpolation algorithms was also discussed, including traditional interpolation algorithm, interpolation based on edge and interpolation based on wavelet domain. The principle of some existing image interpolation algorithms has been introduced. The discussion of the advantage and disadvantages of algorithms has also been done by experiment simulation and the prior studies.Secondly, the analysis of image interpolation algorithm which based on Curvelet transform, simulate and analysis the original algorithm was studied. The coefficient matrix of Curvelet transform was divided into eight directions, which was defined as the direction factor of eight directions at each pixel point, thus, the description of texture and direction is better. The weight coefficients were constructed by the direction factors. So a better algorithm for texture and direction image can be obtained.The analysis of image reconstruction algorithm based on gradient field was studied. Based on this algorithm the edge information of high resolution image will be well kept. However, there are a large of parameters in the algorithm and affect the quality of image. Therefore, the analysis of the influence of various parameters on the results of the original algorithm was studied in this part of work. Optimal parameters were obtained by a lot of tests, and improve the effect of original algorithm. The combination of the iterative back-projection and non-local means enhanced the effect of high resolution image. Compare with the existing algorithms, the results show that the improved algorithm can well preserve edge characteristics, the subjective visual effect and the objective evaluation indicators were all increased.Finally, this work designed the software system of image interpolation algorithm, the application of this system show the effectiveness of the algorithm intuitively in image interpolation technique.