均场退火算法在单幅灰度图像高光检测与恢复中的应用

发布时间：2019-05-12 07:13

【摘要】：为了解决单幅灰度图像高光去除方法恢复结果存在的图像失真问题,提出一种基于均场退火算法的单幅灰度图像高光检测方法.首先利用反射模型分别对镜面反射分量和漫反射分量的分布进行建模;然后通过均场退火算法的迭代过程估计镜面反射分量和漫反射分量的比例,对可能存在的高光区域进行检测;最后利用基于BSCB模型的图像修复方法修复高光区域.采用一种主观评价方法和客观评价方法相结合的性能的评价方法对文中方法进行验证,结果表明,该方法是有效的;与传统的高光检测与恢复的方法相比,该方法能够有效地检测出灰度图像中镜面反射区域,且恢复效果更符合人眼视觉、恢复后的图像质量更好,提高了图像高光区域的恢复率.
[Abstract]:In order to solve the problem of image distortion in the restoration results of single gray image highlight removal method, a single gray image highlight detection method based on mean field annealing algorithm is proposed. Firstly, the distribution of mirror reflection component and diffusion reflection component is modeled by reflection model. Then the ratio of specular component to diffuse component is estimated by the iterative process of the mean field annealing algorithm to detect the possible specular region. Finally the image restoration method based on BSCB model is used to repair the specular region. A performance evaluation method which combines subjective evaluation method with objective evaluation method is used to verify the method in this paper. The results show that the method is effective. Compared with the traditional specular detection and restoration method, this method can effectively detect specular regions in gray-scale images, and the restoration effect is more in line with human vision, and the quality of restored images is better than that of traditional specular detection and restoration methods. The recovery rate of the highlight area of the image is improved.
【作者单位】：哈尔滨理工大学计算机科学与技术学院;哈尔滨理工大学仪器科学与技术博士后科研流动站;哈尔滨工业大学计算机科学与技术学院;
【基金】：国家自然科学基金(61440025) 黑龙江省教育厅科学技术研究项目(12541119)
【分类号】：TP391.41

【相似文献】