IEEE Trans Image Process. 2014 Dec;23(12):5298-308. doi: 10.1109/TIP.2014.2364537.
This paper proposes a novel algorithm, which enhances the contrast of an input image using spatial information of pixels. The algorithm introduces a new method to compute the spatial entropy of pixels using spatial distribution of pixel gray levels. Different than the conventional methods, this algorithm considers the distribution of spatial locations of gray levels of an image instead of gray-level distribution or joint statistics computed from the gray levels of an image. For each gray level, the corresponding spatial distribution is computed using a histogram of spatial locations of all pixels with the same gray level. Entropy measures are calculated from the spatial distributions of gray levels of an image to create a distribution function, which is further mapped to a uniform distribution function to achieve the final contrast enhancement. The method achieves contrast improvement in the case of low-contrast images; however, it does not alter the image if the image’s contrast is high enough. Thus, it always produces visually pleasing results without distortions. Furthermore, this method is combined with transform domain coefficient weighting to achieve both local and global contrast enhancement at the same time. The level of the local contrast enhancement can be controlled. Several experiments on effects of contrast enhancement are performed. Experimental results show that the proposed algorithms produce better or comparable enhanced images than several state-of-the-art algorithms.
本文提出了一种新的算法，利用像素的空间信息来增强输入图像的对比度。该算法引入了一种新的方法来计算像素的空间熵，使用像素灰度级的空间分布。与传统方法不同，该算法考虑了图像灰度级的空间位置分布，而不是从图像的灰度级计算灰度级分布或联合统计。对于每个灰度级，使用所有具有相同灰度级的像素的空间位置的直方图计算相应的空间分布。从图像的灰度级的空间分布中计算出熵度量值，以创建一个分布函数，然后将其映射到均匀分布函数，以实现最终的对比度增强。该方法在低对比度图像的情况下实现了对比度的提高；但是，如果图像的对比度足够高，则不会改变图像。因此，它始终会产生视觉上令人愉悦的结果，而不会产生失真。此外，该方法与变换域系数加权相结合，同时实现局部和全局对比度增强。可以控制局部对比度增强的程度。进行了对比度增强效果的几项实验。实验结果表明，与几种最先进的算法相比，所提出的算法产生了更好或相当的增强图像。
