A new MSBF-based HDR image processing algorithm

High dynamic range (HDR) image is increasingly comprehensive in digital image applications recently. But conventional output equipments and algorithm only provide to support 8 bit image. Here, a new algorithm based on multi-scale bilateral filtering (MSBF) is presented to deal with HDR image, which not only keeps image’s details and colors, but also compresses image’s brightness distributions. Consequently dynamic range compression based on MSR algorithm may be realized in four scales. Wherein, the Gaussian kernel whose scale is small is used to suppress halo, the Gaussian kernel of middle scale and bigger scale is used to keep natural whole colour tune. But the dynamic range compression based on this algorithm just partly weakens local contrast reversal, and can not effectively control the extent of edge diffusion. By edge-cutoff function, bilateral filter may effectively suppress halo, which is formed by local contrast reversal derived from Gaussian filter. In space domain the scale of Gaussian kernel was not sensitive to filtering result when edge-cutoff function based on Gaussian kernel was used, therefore, constant may be chosen as scale of filter. At last, a series of experiments were conducted to test this algorithm. A conclusion can be draw that the MSBF-based algorithm is rather effective in processing HDR image.