光学学报, 2017, 37 (6): 0630003, 网络出版: 2017-06-08   

基于分组三维离散余弦变换字典的植物高光谱数据去噪方法

Denoising Method for Plant Hyperspectral Data Based on Grouped 3D Discrete Cosine Transform Dictionary
作者单位
杭州电子科技大学生命信息与仪器工程学院, 浙江 杭州 310018
摘要
针对植物高光谱图像各波段噪声强度不同, 以及空间域和谱域均存在噪声污染的问题, 提出了一种基于分组三维(3D)离散余弦变换(DCT)字典的稀疏表示去噪方法。首先分析了植物光谱特征, 根据谱间相关性对波段进行分组; 然后采用边缘块剔除的局部均值标准差法对高光谱图像进行噪声标准差估计, 为去噪算法提供参考阈值; 最后构建三维DCT字典的稀疏表示去噪方法, 对植物高光谱图像进行去噪。实验结果表明, 与原始数据和二维DCT字典去噪方法相比, 谱域噪声评估中平均信噪比分别提高18.2 dB和9.2 dB。因此, 该方法不仅具有较好的空间域去噪能力, 也有较好的谱域去噪能力。
Abstract
In order to solve the problems that noise intensity of each band for plant hyperspectral image is different and noise exists in both spatial and spectral domains, a sparse representation denoising method is proposed based on the grouped three-dimensional (3D) discrete cosine transform (DCT) dictionary. Firstly, the spectral characteristics of the plants are analyzed and the bands are grouped according to the spectral correlation. Secondly, local mean standard deviation of eliminating edges is used to estimate the noise standard deviation of hyperspectral images, which provides the reference threshold for denoising algorithm. Finally, a sparse representation denoising method based on 3D DCT dictionary is constructed for denoising plant hyperspectral images. Experimental results show that, comparing with the original data and the denoising method of two-dimensional (2D) DCT dictionary, the average signal-to-noise ratios of the noise evaluation by the proposed method are improved by 18.2 dB and 9.2 dB in the spectral domain. Therefore, the proposed method can denoise not only in the spatial domain but also in the spectral domain.

徐平, 肖冲, 张竞成, 薛凌云. 基于分组三维离散余弦变换字典的植物高光谱数据去噪方法[J]. 光学学报, 2017, 37(6): 0630003. Xu Ping, Xiao Chong, Zhang Jingcheng, Xue Lingyun. Denoising Method for Plant Hyperspectral Data Based on Grouped 3D Discrete Cosine Transform Dictionary[J]. Acta Optica Sinica, 2017, 37(6): 0630003.

本文已被 5 篇论文引用
被引统计数据来源于中国光学期刊网
引用该论文: TXT   |   EndNote

相关论文

加载中...

关于本站 Cookie 的使用提示

中国光学期刊网使用基于 cookie 的技术来更好地为您提供各项服务,点击此处了解我们的隐私策略。 如您需继续使用本网站,请您授权我们使用本地 cookie 来保存部分信息。
全站搜索
您最值得信赖的光电行业旗舰网络服务平台!