Study On Color Filter Array Interpolation And Image Restoration Algorithm

Image will be inevitably affected by the various factors during the process of acquisition,which results in images with varying degrees of degradation,restoring the true image is critical for later image processing.With the popularity of digital cameras,color images emerge in various fields.Although there are some links between color image and grayscale image,however some kind of corss exist within channels of color imge,there will be some sorts of problems if grayscale image restoration algorithm is applied directly to color image.It is difficult both to remove blur while retaining image details for traditional image restoration methods.Introducing the Total Variation Principle to the field of image processing has a very important significance,TV model can remove blur and noises effectively while retaining the edges of image.Total Variation restoring image is obtained by introducing the energy functional,which is converted to the problem of functional extremum problem.This paper focuses on the research of color image acquisition,recovery and objective quality evaluation,mainly from the following three aspects: Firstly,this paper introduces the Bayer(a kind of Color Filter Array,CFA)template interpolation algorithm,namely how to interpolate a color image from the ‘raw' format image obtained from the CCD sensor covering with color filter array.On the basis of analyzing and comparing several traditional CFA interpolation algorithms,a modified iterative interpolation algorithm based on edge type is proposed,experimental results show that this algorithm can effectively suppress the Zipper effect and False Color effect.Based on the Bayer template,the interpolation algorithm of Truesense sparse color filter array which is suitable for weak light condition is studied.Secondly,several classical image restoration models and algorithms are described,and their advantages and disadvantages are analyzed.Extending the grayscale image restoration algorithm based on TV Principle to color image,we focus on the introduction of fast Total Variation Deconvoltion(FTVd)algorithm.This algorithm can be applied to both within channel blur and cross channel blur,and has the advantages of strong convergence,numerical stability and no-sensitive to pathological conditions characteristics.Finally,Quaternion Structural Similarity(QSSIM)image quality evaluation is carried out on the restored color images,which further illustrates the superiority of the TV Principle to restore color images.