基于半像素错位的多幅图像重建高分辨率图像技术研究  

One method of super-resolution image rebuilding using multi-images of half pixel offset

作　　者：黄战华[1] 蔡怀宇[1] 张以谟[1] 李贺桥[1] 夏劲松 李蒙[1] 

机构地区：[1]天津大学精密仪器与光电子工程学院光电信息技术科学教育部重点实验室,天津300072

出　　处：《光学技术》2002年第5期389-391,394,共4页Optical Technique

摘　　要：介绍了一种基于半像素错位的多幅图像重建高分辨率图像技术。分析了半像素错位的多幅图像与高分辨率图像各像素灰度值的对应关系 ,并从CCD数字化采样的角度进行了论证。同时 ,结合实际摄像机CCD结构 ,求出了高分辨率图像重建的计算公式 ,并通过实验进行了验证和完善。重建的本质是以原高分辨率图像的 4邻域平均图像为基础 ,增加一定比例的边缘细节信息 ,去接近原高分辨率图像。CCD的动态范围越大 ,图像的灰度级越多 ,那么计算误差就越小 ,图像的边缘细节信息就可以利用更多 ,重建的图像就越接近原高分辨率图像。通过实验和分析表明 ,利用半像素错位的多幅低分辨率图像重建高分辨率图像的原理是正确的 。One method of super-resolution image rebuilding using multi-images of half pixel offset is introduced. The relation of pixel gray-level of low resolution images and high resolution image is found, and it is demonstrated by analyzing the process of CCD sampling. Correspond to different CCD configuration of camera, the expressions for super-resolution image rebuilding is educed, and it is validated and modified by test. In the process of super-resolution image rebuilding, the smoothed image by adjacent 4 pixels is considered as a start, and it is supplemented with high frequency detail of image, and approachs to the precision super-resolution image. The wider of CCD sensitivity range and pixel gray-levels are, the smaller of computer error occru, then the more high frequency usable detail of image is obtained, and result is more colse to precision super-resolution image. The test and analysis show that the theory of super-resolution image rebuilding using multi-images of half pixel offset is reasonable, and the technique is available.

关 键 词：图像处理 图像重建 超分辨率 像素灰度值误差 CCD 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]