HDR Image Tone Mapping Algorithms Based On Local Edge-Preserving Filtering

In recent years,the high dynamic range image technique which contains colorful details in real scenes becomes popular in computer graphics area because of the natural effeccts which cannot be achieved by the low dynamic range images, especailly the extreme bright and dark counldn`t be observed by naked eyes. Therefore, high dynamic range image have been widely used in the field of professional photography, remote sensing satellite, image rendering, special effects and many other areas. Due to the wide dynamic range of high dynamic range image, which cannot be displayed on low dynamic range device properly, tone mapping algorithm is proposed to compress the dynamic range of the image. The main purpose of tone mapping is compressing the dynamic range of the image to the dynamic range of the display device with the aim to preserve the wealth of detail information and higher overall contrast.The research shows that,most of the tone mapping algorithms can only ensure one aspect of image quality which can result in the poor image contrast, colour distortion, detail losing and flare phenomenon. The tone mapping algorithm is improved from image brightness and color compents which aims at solving the aboving defects. Firstly,the Gaussian Filter function is used to eliminate the nosie and smooth the image. Secondly, the global tone mapping algorithm based on photography is used to compress the luminance of the image which aim at reducing the dynamic range of the image. The image which is processed by local edge-preserving filter algorithm is iteratively devided into a base layer and three detail layers,with the detail layer is mapped by the s-function to enhance the local contrast and the base layer is mapped by the linear function to futher reduce the dynamic range of the image.Then the base layer is combined with the detail layers. Finaly,the image is devided into multiple zones by the k-means clustering method and the gamma correction is received based on the average brightness value of every region, which is used to revise the image.By comparing with other typical of the tone mapping algorithm, the tone mapping algorithms can produce a higher quality image which contains a wealth of detail and higher local contrast with a higher clear and natural image by avoiding the phenomenon of halo.