Abstract: To solve the problem that the 8 bit quantization results of the traditional high dynamic range image compression methods are easily influenced by image background change, quantization images blur, image detail and dim target loss, a novel histogram reconstruction based image detail enhancement algorithm is proposed. Firstly, image histogram statistics is reassigned to save image details and reduce the interval of adjacent gray level. Secondly, Gabor filters are used to simulate visual perception system, and smooth images are acquired by using Gabor filters. Finally, local contrast enhancement method is used to enhance image details, and linear mapping 14 bit enhanced image to 8 bit display image. The experimental results demonstrated that the quantization images of the proposed algorithm are more distinct and have more image details comparing with traditional quantization methods.

Key words: quantization, image detail enhancement, Gabor filters, local contrast enhancement, linear map

摘要： 针对传统大动态范围图像数据压缩方法易受场景变化影响，量化后的8位显示图像整体模糊、图像细节和弱小目标丢失问题，提出了一种基于直方图重建图像细节增强算法。对图像直方图统计值进行重新赋值，保留图像中出现的细节部分，并缩小相邻灰度级间间隔；采用二维Gabor滤波器来模拟视觉感知系统，将Gabor滤波器与图像进行卷积运算，得到滤波后的平滑图像；采用局部对比度增强方法来增强图像细节部分，并将增强后的中间结果线性映射为8位显示图像。实验结果表明，与其他大动态范围数据压缩方法相比，该算法量化后图像清晰度高，无图像细节和目标丢失现象。

关键词: 量化, 图像细节增强, Gabor滤波器, 局部对比度增强, 线性映射