科研之友
微信
新浪微博
Facebook
分享链接
摘要
In this paper, the wavelet packet (WP) decomposition of the hyperspectral image is discussed in spatial and spectral dimensions, and a critical function and an obtaining model of the best decomposition level (BDL) are put forward by the means of a high frequency matrix in the different decomposition levels of the image. In this study, the singular value of decomposition (SVD) coefficient matrix of the critical function was developed after the spatial and spectral hyperspectral image is transformed and decomposed by the WP in the frequency domain, then a regression curve equation and its 2-order derivatives about the critical function values were computed in different decomposition levels in order to acquire the inflection point as a direction of the WP-BDL. In addition, another hyperspectral image is used to test the universality of the critical algorithm and obtaining model. Finally, the hyperspectral images are denoised in the high frequency domain by the means of a soft threshold based on the WP decomposition. The results and their accuracy are better.
- 出版日期2013-7
- 单位中国矿业大学（北京）
