A New Semi-supervised Classification Method of Hyperspectral Image based on Combining Renyi Entropy and Multinomial Logistic Regression Algorithm

Chunyang Wang, Shuangting Wang, Zengzhang Guo, Liping Wang, Chao Ma
2014 International Journal of Signal Processing, Image Processing and Pattern Recognition  
The study on classification methods of hyperspectral image is a focal growing area in remote sensing applications because the wide spectral range, providing a very high spectral resolution, allows the detection and classification surfaces and chemical elements of the observed image. Semi-supervised learning method which takes a large number of unlabeled samples and minority labeled samples, improving classification and predicting the accuracy effectively have been a new research direction. In
more » ... is paper we proposed a new semisupervised classification method of hyperspectral image based on combining Renyi entropy and multinomial logistic regression algorithm. The multinomial logistic regression was performed to describe a direct relationship between the selected sample as and their category. A lot of unlabeled samples are constantly added to the sample data using Renyi entropy algorithm. The test analysis of image classification in test area showed the advantages of classification method based on combining Renyi entropy and multinomial logistic regression algorithm for hyperspectral remote sensing image. maximum information in hyperspectral image to add to the training samples dataset. Finally, the proposed method iterates the selection and learning processes to output the final classification result. The classification test of study area image was conducted using the new algorithm. The experimental results showed that the algorithm has a high precision overall classification, less misclassification errors and leakage points, and more advantages at intensive more multiclass region classification. However, the new algorithm still has inadequacies, such as long-running time and need further more computing power and hardware. Future research priority will focus on optimizing algorithm, saving the running time, and promoting working efficiency. Authors Chunyang Wang, he received his B. Eng. degree from Changchun . His research interests include image processing, classification method of hyperspectral image, machine learning and pattern recognition of remote sensing application. Shuangting Wang, he is a full Professor in School of Surveying and Land Information Engineering, Henan Polytechnic University, Jiaozuo, China. His research interests include photogrammetry, remote sensing image processing, remote sensing information extraction. Zengzhang Guo, he is a full Professor in
doi:10.14257/ijsip.2014.7.5.06 fatcat:nbajzsn4engv5njy3jxsnl6sge