A Genial Method of Color Features Extraction In Remote Sensing Images Using Multi Kernel Principal Component Analysis With Multi-Level 2-D Wavelet Construction

One of the most significant requirements in image retrieval, classification, clustering and etc. is extracting efficient features from an image. The color feature is one of the visual features. In this proposed work, various features in a remote sensing image can be distinguished based on their color. The features are extracted as object and distinguished with color. Initially, Gaussian noise is added to image and multi-level 2-D wavelet construction is applied to get denoised image. Next, proposed Multi Kernel Principal Component Analysis preserves local and global structure of data sets and also handles heterogeneous characteristics in an image. Finally, Fuzzy C-Means clustering algorithm partitions the datasets into clusters so that data in each cluster shares some common characteristic which is integrated with color conversion method to extract feature based on color present in the image. The performance of this proposed work is measured through various performance metrics to analysis best result for feature extraction of remote sensing images.