基于双直方图均衡算法的红外图像增强

为了抑制全局直方图均衡产生的灰度饱和和局部细节丢失的情况，提出了一种双直方图均衡算法。首先对图像的背景和前景进行分割，提出基于直方图的局部最小值和修正的K-Means聚类算法来确定图像的理想分割阈值，然后再对分割的子图分别作全局直方图均衡（Global Histogram Equalization，GHE）。对该算法进行了实验验证，结果表明，相较于GHE算法，经该算法增强后的图像峰值信噪比（Peak Signal to Noise Ratio，PSNR）提高约16.425%，结构相似度（Structural Similarity Index，SSIM）提高约14.85%。同时通过主观分析，基于直方图局部最小值和修正的K-Means聚类算法的图像分割进行双直方图均衡可以有效抑制GHE算法产生的灰度饱和和细节丢失现象。

Bi-Histogram Equalization Algorithm for Infrared Image Enhancement

A modified bi-histogram equalization algorithm is proposed to suppress gray saturation and loss of local details caused by global histogram equalization. First, the background and foreground of the image are segmented, and a modified k-means clustering algorithm based on the local minimum of the histogram is proposed to determine the ideal segmentation threshold of the image. Then, histogram equalization is performed for the segmented sub-graphs. The algorithm is verified by experiments; the results for the experiment show that, compared with those from global histogram equalization, the peak signal to noise ratio and structural similarity are improved by approximately 16.425% and 14.85%, respectively. Simultaneously, through subjective evaluation, the algorithm based on histogram local minimum and modified k-means can effectively suppress the gray saturation and detail loss caused by GHE.