Bayer Image Compression And Interpolation Based On Wavelet Transform And Edge Detection

In recent years, the popularity of digital cameras is increasing,which has influenced on all aspects of research and life.Bayer filter array is the most commonly used filter arrays for a single CCD sensor digital camera. With the resolution of current image sensors becoming more and more, requirements for storage space and transmission bandwidth also are improving rapidly. For the traditional way of compression ,which should be interpolated first, the amount of data is 3 times of the raw data directly compressed.Therefore,raw data compression on the Bayer filter array data is of great significance.Bayer filter array, each pixel data only retain a color, need to get full color pictures making use of interpolation, the interpolation algorithm has the necessity and reality.The work of this paper consists of two parts:(1)Part of the original data compressionFirst, the wavelet compression algorithm is studied.This paper analyzed and compared the previous algorithm ,in this foundation, selected performed better SPIHT coding as the core of the original data compression algorithm.According to features that Bayer filter array should not be compressed directly, this paper improved two methods of structural transformation for raw data compression,which based on the analysis and comparison of the predecessors of the Bayer filter array data compression algorithm , and combined with wavelet lifting and SPIHT compression algorithm .After testing,the compression algorithms of this paper is better than the original compression algorithm, the overall compressed effect is better. A modified SPIHT algorithm with practical value is added to this algorithm as a backup option. This algorithm is applied to image that some region of interest is of great importance but background is of no use.(2) Part of CFA interpolationIn this paper, the CFA interpolation algorithms of these years are studied and classified. A few typical interpolation algorithm are highlighted, and their advantages and disadvantages of each are analysed.After analysis distortion of the CFA interpolation algorithm, the corresponding responses are introduced. CFA interpolation algorithms always cause the distortion of edge, while the existing algorithms are difficult to get a better interpolation results.A adaptive approximation interpolation algorithm based on the type of the edges is explored. After forecasting the type of the edges , different adaptive interpolation steps are applied according to the type of the edge. Experimental results show that the algorithm can take into account edge interpolation effects for various edges.To improve the interpolation quality, in the post-processing step, iterative interpolation step are applied. Benefit by using the low-complexity constant color-difference interpolation step, and a reasonable terminating conditon of iteration, this algorithm could keep the number of iterations about as many as twice.Terefore,this process could achieve high-quality performance while maintaining low complexity. In the last part of the interpolation algorithm,blocking image interpolation step is explored as a backup option, and complexity of control is presented for improving the availability of this algorithm.