• Dr.D.Nagaraju
• G.Prabhakar

To analyze spatial information for hyperspectral image (HSI) classification, CNN-based methods typically crop pixels along with their adjacent pixels from hyperspectral data to form HSI cubes. However, these cropped HSI cubes often have complex spatial land-cover distributions, and the land-cover label of a cropped HSI cube cannot be determined solely by its center pixel. Additionally, the spatial land-cover distribution within a cropped HSI cube is fixed and lacks diversity. Consequently, training with these HSI cubes leads to poor generalization in response to variations in spatial land-cover distributions. In this paper, we propose an end-to-end fully convolutional segmentation network (FCSN) to simultaneously identify the land-cover labels of all pixels within an HSI cube. First, we conduct several experiments to demonstrate the weak generalization capabilities of recent CNN-based methods. Second, we introduce a fine labeling method that assigns detailed spatial land-cover labels to all pixels in HSI cubes. Third, we present a new method for generating HSI cubes with fine labels, enhancing the diversity of spatial land-cover distributions. Finally, the FCSN is designed to explore spectral-spatial features from these finely labeled HSI cubes, improving HSI classification.

Component,formatting,style,styling,insert

"A MapReduce Based Approach for Secure Batch Satellite Image Encryption", International Journal of Emerging Technologies and Innovative Research (www.jetir.org), ISSN:2349-5162, Vol.11, Issue 6, page no.570-575, June-2024, Available :http://www.jetir.org/papers/JETIRGL06095.pdf

"A MapReduce Based Approach for Secure Batch Satellite Image Encryption", International Journal of Emerging Technologies and Innovative Research (www.jetir.org | UGC and issn Approved), ISSN:2349-5162, Vol.11, Issue 6, page no. pp570-575, June-2024, Available at : http://www.jetir.org/papers/JETIRGL06095.pdf