针对高动态范围(High Dynamic Range, HDR)图像的动态范围与普通显示设备不匹配的问题, 本文提出了一种亮度分区和导向滤波相结合的色调映射算法。首先, 通过颜色空间转换将输入的HDR图像转换到CIEL*a*b*空间, 得到亮度层和色度层图像, 接着对亮度层图像进行对数压缩并分区, 对压缩后的各分区亮度图像选取合适γ值进行校正。然后, 对于现有映射算法过于强调对图像亮度的处理而忽略色度的问题, 本文将压缩后的亮度图像作为引导图像对色度层图像进行导向滤波处理。最后, 将处理后的亮度层图像和色度层图像融合得到标准动态范围(Standard Dynamic Range, SDR)图像。通过对比实验表明, 本文算法得到的图像在清晰度、信息熵、方差方面较全局映射法、局部映射法和双边滤波法平均提高了94.03%、30.73%和24.23%, 说明本文算法在实现高动态图像亮度压缩的同时, 更好地实现了对色度和局部细节的处理。

Aiming at the problem that the dynamic range of high dynamic range (HDR) image does not match the ordinary display device, a tone mapping algorithm combining luminance partitioning and guidance filtering is proposed. Firstly, the input HDR image is converted into the CIEL*a*b* space by color space conversion to obtain the luminance layer and the chrominance layer image. Then the luminance layer image is logarithmically compressed and luminance partitioned, and the compressed partition is performed. The luminance image is corrected by selecting the appropriate γ value. Then, for the problem that the existing mapping algorithm to overemphasize the processing of image luminance and ignore the chromaticity, the compressed luminance image is used as the guiding image to perform the guiding filtering processing on the chrominance layer image. Finally, the processed luminance layer image and the chrominance layer image are fused to obtain a standard dynamic range (SDR) image. The comparison experiments show that the images obtained by the proposed algorithm are 94.03%, 30.73% and 24.23%, higher than the global mapping method, local mapping method and bilateral filters method in terms of resolution, information entropy and variance. This paper proves that the algorithm achieves high dynamic image luminance compression. At the same time, the processing of chromaticity and local detail are better realized.