基于直方图分析的航空摄像机自动调光方法研究  

Auto Exposure Algorithm for Aerial Camera Based on Histogram Statistics Method

作　　者：申小禾 刘晶红[1] 楚广生[1] 

机构地区：[1]中国科学院长春光学精密机械与物理研究所,长春130033 [2]中国科学院大学,北京100049

出　　处：《电子与信息学报》2016年第3期541-548,共8页Journal of Electronics & Information Technology

基　　金：国家863高科技研究发展计划(2008AA121803);中国科学院航空光学成像与测量重点实验室基金(2012MS06)~~

摘　　要：航空摄像机在空中对地面进行拍摄,拍摄场景复杂,传统的灰度平均值法在一些如有干扰点或背景与目标灰度差别大的场景下曝光会不正确,丢失图像信息,为了解决这些问题,该文在根据图像灰度信息的调光算法以及硬件实现方面进行了相关研究。首先根据当前帧图像的灰度直方图中灰度级为0和255附近的信息进行初调;然后根据图像直方图及其卷积和,搜索其中的峰值区域并确定图像的不感兴趣区域(包含的灰度级单一,信息量少的区域);接着根据这些区域占整幅图像的比重,由权值分配曲线为其分配不同的权值;最后计算加权后的灰度平均值,并将其作为反馈量控制下一帧的调光。实验结果表明,在普通场景下能够准确曝光,在暗背景和亮背景场景下,相比传统方法使图像感兴趣区域信息熵提高10%以上。此方法能适用于多种场景,满足航空摄像机对复杂场景自动调光的需求。The aerial camera takes photographs from sky. Conventional auto exposure algorithms can not be suitable for some special scenes such as the background is bright or dark and few strong interference points exist. Unsuitable exposure causes information losing in images. In order to solve this problem, an algorithm based on gray value is proposed. Firstly, it regulates the exposure preliminarily using the information of the histogram near 0 and 255. Then it obtains the regions of no interest in the image through the convolution of the histogram and weighted different proportion for the regions according to their area in the entire image. Finally, the weighted average is computed, which is regarded as the feedback controlling exposure. Experimental results indicate that image entropy of regions of interest increased more than 10% in dark or bright background. This method is suitable for many scenes, and meets the demands of the aerial camera.

关 键 词：图像处理 航空摄像机 灰度直方图 加权平均值 自动调光 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]