改进匹配块的Criminisi修复算法
|
摘要
图像修复技术在图像处理中有着非常重要的作用,它在图像拼接,图像编辑和目标移除等方面有着广泛的应用。针对原始Criminisi算法中图像修复之后出现图像修复效果不佳的问题提出了一种改进的Criminisi算法。改进算法针对原始Criminisi算法中修复块大小固定,提出一种利用峰值信噪比来寻找最佳修复块。通过大量实验验证,改进的算法能够有效解决图像修复中的连续性问题,提高修复的质量与效率。
Image restoration technology plays a very important role in image processing. It has a wide range of applications in image mosaic, image editing and object removal. An improved Criminisi algorithm is proposed to solve the problem of poor image restora‐tion effect after image restoration in the original Criminisi algorithm. Aiming at the fixed size of the repair block in the original Cri‐minisi algorithm, the improved algorithm proposes a peak signal-to-noise ratio(PSNR) to find the best repair block. Experiments show that the improved algorithm can effectively solve the continuity problem in image restoration and improve the quality and effi‐ciency of image restoration.
Image restoration technology plays a very important role in image processing. It has a wide range of applications in image mosaic, image editing and object removal. An improved Criminisi algorithm is proposed to solve the problem of poor image restora‐tion effect after image restoration in the original Criminisi algorithm. Aiming at the fixed size of the repair block in the original Cri‐minisi algorithm, the improved algorithm proposes a peak signal-to-noise ratio(PSNR) to find the best repair block. Experiments show that the improved algorithm can effectively solve the continuity problem in image restoration and improve the quality and effi‐ciency of image restoration.
引文
[1]陈钱.红外图像处理技术现状及发展趋势[J].红外技术,2013, 35(6):311-318.
[2] Bertalmío M, Sapiro G, Caselles V, et al. Image inpainting[C]//Conference on Computer Graphics and Interactive Techniques.DBLP, 2000:417-424.
[3] Bertalmio M, Vese L, Sapiro G, et al. Simultaneous structureand texture image inpainting.[J]. IEEE Transactions on ImageProcessing, 2003, 12(8):882-889.
[4] Shen J. and Chan T. Mathematical Models for Local Non-tex‐ture Inpaintings.[J]. Siam Journal on Applied Mathematics,2002, 62(3):1019-1043.
[5] Criminisi A, Pérez P, Toyama K. Region filling and object re‐moval by exemplar-based image inpainting[J].IEEE Transac‐tions on Image Processing, 2004, 13(9):1200-1212.
[6]池悦,何宁,张琪,等.一种改进的基于Criminisi算法的目标移除方法[J].北京联合大学学报, 2017, 31(1):67-74.
[7]李爱菊,钮文良.基于改进Criminisi算法的图像修复[J].计算机工程与应用, 2014, 50(18):167-170.
[8]张申华,王克刚,祝轩.局部特征信息约束的改进Criminisi算法[J].计算机工程与应用, 2014, 50(8):127-130.
[9]汪方正,吴建鑫.基于纹理结构引导自适应的改进Criminisi算法[J].武汉科技大学学报, 2016, 39(2):150-155.
[10]李尊,吴谨,刘劲.目标移除的Criminisi图像修复算法[J].红外技术, 2016, 38(1):28-32.
[11]李尊,吴谨,刘劲.数学形态学在Criminisi图像修复算法中的应用[J].红外技术, 2015, 37(7):574-578.
[12]刘奎,苏本跃,赵晓静.基于结构张量的图像修复方法[J].计算机应用, 2011, 31(10):2711-2713.
[13] Lore K G, Akintayo A, Sarkar S. LLNet:A deep autoencoderapproach to natural low-light im-age enhancement[J]. PatternRecognition, 2017, 61:650-662.
[2] Bertalmío M, Sapiro G, Caselles V, et al. Image inpainting[C]//Conference on Computer Graphics and Interactive Techniques.DBLP, 2000:417-424.
[3] Bertalmio M, Vese L, Sapiro G, et al. Simultaneous structureand texture image inpainting.[J]. IEEE Transactions on ImageProcessing, 2003, 12(8):882-889.
[4] Shen J. and Chan T. Mathematical Models for Local Non-tex‐ture Inpaintings.[J]. Siam Journal on Applied Mathematics,2002, 62(3):1019-1043.
[5] Criminisi A, Pérez P, Toyama K. Region filling and object re‐moval by exemplar-based image inpainting[J].IEEE Transac‐tions on Image Processing, 2004, 13(9):1200-1212.
[6]池悦,何宁,张琪,等.一种改进的基于Criminisi算法的目标移除方法[J].北京联合大学学报, 2017, 31(1):67-74.
[7]李爱菊,钮文良.基于改进Criminisi算法的图像修复[J].计算机工程与应用, 2014, 50(18):167-170.
[8]张申华,王克刚,祝轩.局部特征信息约束的改进Criminisi算法[J].计算机工程与应用, 2014, 50(8):127-130.
[9]汪方正,吴建鑫.基于纹理结构引导自适应的改进Criminisi算法[J].武汉科技大学学报, 2016, 39(2):150-155.
[10]李尊,吴谨,刘劲.目标移除的Criminisi图像修复算法[J].红外技术, 2016, 38(1):28-32.
[11]李尊,吴谨,刘劲.数学形态学在Criminisi图像修复算法中的应用[J].红外技术, 2015, 37(7):574-578.
[12]刘奎,苏本跃,赵晓静.基于结构张量的图像修复方法[J].计算机应用, 2011, 31(10):2711-2713.
[13] Lore K G, Akintayo A, Sarkar S. LLNet:A deep autoencoderapproach to natural low-light im-age enhancement[J]. PatternRecognition, 2017, 61:650-662.