期刊名称：International Journal of Advanced Research in Computer Engineering & Technology (IJARCET)
印刷版ISSN：2278-1323
出版年度：2015
卷号：4
期号：7
页码：3312-3318
出版社：Shri Pannalal Research Institute of Technolgy
摘要：Hyper spectral images are those where each pixel forms an almost continuous spectrum. They have experienced significant success but in practice they are degraded by a mixture of various types of noises i.e. Gaussian noise, dead pixels or lines, stripes and so on. A Hyper Spectral Image restoration method is introduced which is based on low-rank matrix recovery (LRMR) and Neural Network which remove the Gaussian noise, dead pixels or lines and stripes. This algorithm is applied on different size of images having different spectral bands. In order to recover the missing pixels or neighbouring pixels, connected component analysis with indexing is used. This paper proposes image restoration of hyper spectral images using LRMR and Neural network which promise qualitative and quantitative result of the degraded images in terms of Peak Signal to Noise Ratio, Mean Square Error, Bit Error Rate and Structural Similarity Index.
