﻿ 基于梯度L0范数的敦煌壁画图像平滑 Dunhuang Murals’ Image Smoothing Processing Research Based on Gradient L0 Norm

Computer Science and Application
Vol.06 No.07(2016), Article ID:18097,6 pages
10.12677/CSA.2016.67048

Dunhuang Murals’ Image Smoothing Processing Research Based on Gradient L0 Norm

Hongqiang Zhang1, Chunxi Wang2, Bingkun Pi1, Cong Ma1, Shuwen Wang1*

1Northwest University for Nationalities, Lanzhou Gansu

2Zhengzhou Meteorological Bureau, Zhengzhou Henan

Received: Jul. 16th, 2016; accepted: Jul. 24th, 2016; published: Jul. 27th, 2016

ABSTRACT

This paper aims to achieve a new way to display Dunhuang murals, and by using gradient L0 norm smooth Dunhuang murals, researches the best iteration times for Dunhuang murals to obtain the best smoothing results. By this way in the new show works, we can ignore the damaged part and clear its characters and scene at the same time. And we will fully demonstrate its significance for others.

Keywords:Dunhuang Murals, Image Smoothing, Gradient L0 Norm

1西北民族大学，甘肃 兰州

2郑州市气象局，河南 郑州

1. 引言

(1) 通过去除小的非零梯度，抚平不重要的细节信息；

(2) 增强图像显著性边缘。

2. L0平滑

L0平滑是香港中文大学徐立等人提出的一种保留边界的图像平滑方法 [5] [6] ，是一种基于图像梯度零范数的全局平滑方法。其主要思想是：通过是的图像梯度的零范数更小来实现图像的平滑处理，目标函数如公式(1)：

(1)

(2)

L0范数 [7] [8] 可以理解为向量中非零元素的个数。图像梯度L0范数可以如公式(3)表示：

(3)

(4)

3. 敦煌壁画的平滑处理及结果分析

(a) (b)(c) (d)

Figure 1. The Dunhuang murals’ image smoothing results 1. (a) is the original, (b) is the smooth results when the number of iterations is 5 times, (c) is the smooth results when the number of iterations is10 times, (d) is the smooth results when the number of iterations is 20 times

(a) (b)(c) (d)

Figure 2. The Dunhuang murals’ image smoothing results 2. (a) is the original, (b) is the smooth results when the number of iterations is 5 times, (c) is the smooth results when the number of iterations is 10 times, (d) is the smooth results when the number of iterations is 20 times

(a) (b)(c) (d)

Figure 3. The Dunhuang murals’ image smoothing results 3. (a) is the original, (b) is the smooth results when the number of iterations is 5 times, (c) is the smooth results when the number of iterations is 10 times, (d) is the smooth results when the number of iterations is 20 times

(a) (b)

Figure 4. The Dunhuang murals’ image smoothing results 4. (a) is the original, (b) is the smooth results when the number of iterations is 5 times

(a) (b) (c)

Figure 5. The smoothing results of the traditional algorithm

(a) (b)

Figure 6. Comparison of two methods for smoothing results. (a) is the smoothing result of the neighborhood averaging method. (b) is the smoothing result of the L0 smoothing

4. 结语

Dunhuang Murals’ Image Smoothing Processing Research Based on Gradient L0 Norm[J]. 计算机科学与应用, 2016, 06(07): 393-398. http://dx.doi.org/10.12677/CSA.2016.67048

1. 1. 王书文, 骆岩红, 黄伟, 杨筱平, 贾建芳. 敦煌壁画数字图像修复中遇到的挑战[J]. 西北民族大学学报(自然科学版), 2009, 30(2): 44.

2. 2. Xu, L., Lu, C.W., Xu, Y., et al. (2011) Image Smoothing via L0 Gradient Minimization. ACM Transactions on Graphics (TOG), 30, 174. http://dx.doi.org/10.1145/2070781.2024208

3. 3. Lu, C.W., Xu, L. and Jia, J.Y. (2012) Combine Sketch and Tone for Pencil Drawing Production. Proceedings of the Symposium on Non-Photorealistic Animation and Rending, Annecy, 65-73.

4. 4. Kang, H., Lee, S. and Chui, C.K. (2007) Coherent Line Drawing. Proceedings of the 5th International Symposium on Non-Photorealistic Animation and Rendering, San Diego, 43-50. http://dx.doi.org/10.1145/1274871.1274878

5. 5. 李效珍, 钱卫东, 叶建亚, 李婷, 赵振兵. 基于Lo平滑和随机游走的变电设备红外图像分割[J]. 激光与红外, 2014, 44(7): 810-814.

6. 6. 刘磊, 陈越, 盛蕴, 张桂戌. 基于自适应噪声模型和线积分卷积的铅笔画模拟[J]. 图学学报, 2015, 36(1): 78.

7. 7. 曾军英, 甘俊英, 翟懿奎. Gabor字典及L0范数快速稀疏表示的人脸识别算法[J]. 信号处理, 2013, 29(2): 256- 161.

8. 8. 王峰, 向新, 易克初, 熊磊. L0范数平滑逼近的稳健求解算法[J]. 电子与信息学报, 2015, 37(10): 2377-2382.