增强景象匹配鲁棒性的灰度变换技术研究

Researches Into Gray Value Transform to Improve Scene Matching Robustness

高可靠高精度的景象匹配技术是自主导航飞行器图像制导技术的核心和关键所在,文章对匹配图像的灰度畸变进行了研究,提出了基于直方图相似变换灰度畸变校正技术,提高了景象匹配定位的准确性和可靠性。由于外部成像条件和成像传感器本身存在差异,匹配图像往往存在较大灰度畸变,使得实时图和基准图在灰度域存在较大不一致,从而影响景象匹配可靠性。文章通过对匹配图像进行直方图的相似变换,将匹配图像直方图调为一致,有效消除了匹配图像灰度差异,调整后图像间的直方图相似系数均在0.95以上。通过10300次独立真实外场景象匹配定位试验,证明本文基于直方图的匹配图像灰度预处理方法可有效消除图像间的灰度差异,减少匹配图像间由于灰度差异而产生的重复模式,大幅提高景象匹配鲁棒性,经过本文方法预处理后的匹配图像景象匹配成功率为99.98%,已满足工程实用对景象匹配算法高可靠性的要求。

Scene match arithmetic which has high dependability and precision is the most important factor of image navigation for independent navigation aircraft. The current study has researched into gray distortion of scene match images,and brought forward the emendation method based on the histogram similar transform which can improve the veracity and reliability of scene matching arithmetic. For different external condition and imaging apparatus,there are many gray differences in images participating in scene match,which will affect the scene match robustness seriously. Via adjusting the histogram of scene match images to make them identical,this research eliminated the gray distortion. As a result,the scene match images histogram similar coefficients were all more than 0.95. 10 300 true outfield scene matching orientation experimentations have testified that the pretreatment method for gray images based on histogram can effectively eliminate the gray distortion and decrease the repetitious patterns caused by gray distortion. The success ratio of scene match has reached the level of 99.98%,which has met the high dependability need of projects.