为了解决在距离选通型水下激光成像系统图像增强中直方图均衡所带来的灰度级过多简并、背景提升的问题, 研究了改进的基于直方图的图像增强技术。首先研究最优化问题, 提出了双种群的量子粒子群算法, 并以此求解图像加权对比度熵最大化意义下最优平台直方图均衡化。接着提出了改进的直方图投影算法, 以二维直方图提取灰度级的相关性信息, 并设计两个权值, 得到加权后的直方图。实验结果表明, 双种群量子粒子群算法具有较好的全局收敛性, 而改进的直方图投影算法在抑制灰度级简并和增加增强效果方面进行了更好的折中。

In order to solve the problem of gray-level combination and background promotion owing to histogram equalization in image enhancement for range-gated underwater laser imaging system, the improved enhancement algorithm based on histogram is studied in this paper. Firstly, the optimization is considered. The double populations BQPSO is proposed to find the optimal plateau in terms of weighted contrast entropy. Secondly, in order to implement it by hardware expediently, an improved histogram projection algorithm is proposed, which collects the relativity of pixels with 2-D histogram and designs two weighs to amend the original projected histogram. Experiment shows that the proposed double populations BQPSO has good global convergence and the improved histogram projection algorithm has good compromise between restraining the gray-level coalition and increasing intensity of enhancement.