红外图像实时增强的新算法  

New real-time enhancement algorithm for infrared images

作　　者：王炳健[1] 刘上乾[1] 拜丽萍[1] 

机构地区：[1]西安电子科技大学技术物理学院,陕西西安710071

出　　处：《光电工程》2006年第1期46-49,共4页Opto-Electronic Engineering

基　　金：国家自然科学基金资助项目(60377034)

摘　　要：针对红外图像的特点,提出了一种红外图像实时增强的新算法。该算法通过分析图像的直方图,得到图像中目标像素数峰值的估计值,并作为平台直方图均衡化的阈值。用该阈值对直方图进行修正,然后通过修正后的直方图进行直方图均衡化。在FPGA内通过采用并行处理结构及流水线技术实现了该算法,并且每秒可处理25帧128×128×8bits的红外图像。理论分析和实验结果均表明,本算法克服了采用一般直方图均衡化增强红外图像的缺点?对背景和噪声增强过度,抑制了目标的增强。该算法对红外图像增强后,图像对比度是直方图均衡化增强后图像对比度的1.8倍。A new real-time contrast enhancement algorithm for infrared images is presented. By analyzing the histogram of image and self-adaptively selecting threshold value, the modified histogram is gotten. This algorithm can greatly enhance the contrast of infrared images by histogram equalization using the modified histogram. It has advantages of enhancing targets and suppressing backgrounds over general histogram equalization. The contrast of images enhanced by this algorithm is improved approximately 1.8 times than that of images enhanced by general histogram equalization. And it can be easily implemented by in FPGA for real-time imaging applications. This paper describes a simple and effective implementation of this algorithm by using pipeline and parallel computation architecture. It can process 25 frames images of 128× 128×Sbits per second.

关 键 词：红外图像 图像增强 直方图均衡化 FPGA 

分 类 号：TN911.73[电子电信—通信与信息系统]