Abstract:Supplementing spectral information with spatial information to improve the classification of hyperspectral image is becoming a hot research in recent years.An improved scheme was put forward according to existing methods.An algorithm of supervised classification was proposed which was combined with bilateral filter and pixel neighborhood information (BS-SVM).Firstly,the spatial texture information of hyperspectral image was extracted whose dimensionality was reduced by PCA.Secondly,spatial correlation information was formed by building pixel neighborhood information of hyperspectral image.Finally,spatial-spectral information was merged by the two kinds of spatial information and the spectral information,which was classified by SVM.The BS-SVM classification method was implemented on the hyperspectral data of Indian Pines and Pavia.The results indicated that in the first place,the OA (Overall accuracy) of G-SVM for Indian Pines and Pavia were 3％～4％ and 2％～3％ higher than those of SVM,the same index for B-SVM were 3％～4％ higher than that of GSVM,and the classification performance can be improved effectively by the spatial texture information of hyperspectral image extracted by bilateral filter.Furthermore,the salt and pepper can be removed effectively by BS-SVM,showing very good performance in hyperspectral classification.In the second place,the classification of some methods for Pavia was better than the Indian.The reason was that the types and distribution of grounds for Indian were more complicated than Pavia.The classification for the less ground were bad,especially the Oats (only 20) was the worst.Therefore,it directly led to the AA (Average accuracy) generally lower than OA.However,the standard deviation of the classification for BS-SVM was much smaller than those of other methods,and the effectiveness of the method was verified with good stability.The experiments showed that the BS-SVM algorithm was better than original SVM with the pure spectrum information,the spatial-spectral information-based methods with Gabor.With the spatial correlation information extracted by the bilateral filter and the pixels neighborhood information,the performance of the classification with BS-SVM algorithm was greatly improved,and the effectiveness of BS-SVM was fully verified in the classification of hyperspectral image.The method can be applied to the field of crop growing,accurate classification and identification.

Image Spectral Data Classification Using Pixel-Purity Kernel Graph Cuts And Support Vector Machines: A Case Study Of Vegetation Identification In Indian Pine Experimental Area

Kernel-like Impurity Detection According to Colour Band Spectral Image Using GA/SVM

A Spectral-Texture Kernel-Based Classification Method for Hyperspectral Images.

A Method of Particle Swarm Optimized SVM Hyper-spectral Remote Sensing Image Classification

An accurate SVM-based classification approach for hyperspectral image classification

Spectral-Spatial Classification of Hyperspectral Images Using Joint Bilateral Filter and Graph Cut Based Model.

Hyperspectral Image Classification Method Combined with Bilateral Filtering and Pixel Neighborhood Information

Hyperspectral Image Classification Based on Iterative Support Vector Machine by Integrating Spatial-Spectral Information

PCA-based Feature Reduction for Hyperspectral Remote Sensing Image Classification

Spectral and Texture Features Combined for Forest Tree Species Classification with Airborne Hyperspectral Imagery

Hyperspectral Image Refined Plant Classification By Graph_based Composite Kernel

Hyperspectral image classification based on spectral-spatial combination features and graph cut

Forest tree species classification based on airborne hyperspectral imagery

An Approach for Hyperspectral Image Classification by Optimizing SVM Using Self Organizing Map

Hyperspectral Image Classification with Combination of Sparse Characteristic and Neighborhood Similarity Metrics

Hyperspectral Image Classification Based on Pre-Post Combination Process.

A SVM ensemble approach combining pixel-based and object-based features for the classification of high resolution remotely sensed imagery

Kernel Extreme Learning Machine Optimized by the Sparrow Search Algorithm for Hyperspectral Image Classification

Spectral-spatial based super pixel remote sensing image classification

A Hyperspectral Image Classification Method Using Multifeature Vectors and Optimized KELM

Classification of urban vegetation patterns from hyperspectral imagery: Hybrid algorithm based on genetic algorithm tuned fuzzy support vector machine