Study On Multi-scale Decomposition Based Tone Mapping Algorithm

With the development of computer technology and the continuous improvement of people's quality of life,different fields including medical images and High Definition movies have put forward higher requirements on image quality.The reason of high dynamic range images becoming a hot research is that it almost contains all data in real scene.The ordinary display device is based on eight bit storage,can only display a low dynamic range image,but can not fully display the high dynamic range image.Aiming at mismatching dynamic range,this thesis deeply studies the algorithm mapping the high dynamic range image into low dynamic range image,and it follows two aspects:A tone mapping algorithm based on multi-scale Improved Local Extrema(ILE)filter is proposed,which has the advantages of high efficiency.This algorithm improves the traditional Local Extrema(LE)filter.The shortcoming of time-consuming is overcame by computing the envelope in the correlation of pixels around the extreme point.The algorithm flow is as follows:(1)To get the basic image and detail images of different scales,the original image is decomposed by multi-scale ILE filter.(2)The fusion image is obtained by merging the basics and details.(3)The final result is obtained by the chroma processing and gamma correction.The result has shown that the efficiency of the algorithm is far higher than other classical tone mapping algorithms such as LE.In subjective evaluation,it can effectively display the details of the image,color and brightness.The objective evaluation shows the local structure of fidelity,natural degree in enhanced images can be preserved well and the result can maintain a high image quality score.On the basis of the above algorithm,an improved multi-scale decomposition framework and fusion algorithm are proposed.To improve the principle of low dynamic range fusion,the advantages of multisclae decomposition and reference image quality evaluation are well exhibited in the improved algorithm.More significant features are included in basic images and detail images.The algorithm flow is as follows:(1)To get basic images and detail images of different scales,improved multi-scale decomposition framework is used to decompose the gray.(2)Basic images are obtained by merging saliency features.(3)The fusion image is obtained by the basic image and detailed images with the detailed features.(4)The final result is obtained by the chroma processing and gamma correction.The experimental results indicate that,under the conditions of acceptable time-consuming,the details of the edge and texture are well-preserved in from subjective evaluation.In objective evaluation,the higher quality evaluation score can be obtained.