基于深度信息的图像修复算法imagecompletionalgorithmbasedon.pdfVIP

Journal of Computer Applications ISSN 1001-9081 2015- 【-10 计算机应用,2015, 【( 【): 【- 【 CODEN JYIIDU 基于深度信息的图像修复算法 1 1 1 1 1 1 何埜 ，李光耀 ，肖莽 ，谢力 ，彭磊 ，唐可 (1. 同济大学 电子与信息工程学院，上海 201804) (*通信作者电子邮箱 saiheye2009@163.com) 摘 要: 针对图像修复结果中存在物体结构上的不连续性和不完整性的问题，提出一种基于深度信息的图像修复算法。首 先，通过建立平面参数马尔科夫模型来推测图像中像素点所在场景空间中的深度信息，从而划分出图像中的共面区域，定位 匹配块的目标位置；其次，根据透视投影原理推导出相应的变换矩阵，指导对匹配块的几何变换；最后，设计含义深度项的 目标代价函数，通过全局优化选择最优修复效果。实验的主观效果对比和PSNR 数据都表明该方法更为优越。 关键词: 图像修复；深度信息；透视投影；全局优化 中图分类号: TP391.41 文献标志码: A Image completion algorithm based on depth information 1 1 1 1 1 1 HE Ye , LI Guangyao , XIAO Mang , XIE Li , PENG Lei , TANG Ke (1.College of Electronics and Information Engineering, Tongji University, Shanghai 201804, China) Abstract: Aiming at the issue that occur in object structure discontinuity and incompleteness in image completion, an image completion algorithm based on depth information is proposed. Firstly, establish the Plane Parameter Markov Random Field model to speculate depth information of the pixels in the image where the scene situate, which are divided into coplanar region in the image, and locate the target matching blocks; Secondly, according to the principle of perspective projection, the transformation matrix is derived, that guides the geometric transformation of the matching block. Finally, the target cost function which include the depth term is designed. Experimental results show the proposed algorithm has superiority in both subjective details and PNSR statistics. Keywords: image completion; depth inforpmation; perspective projection; global optimization