Skip to Main Content
This article introduces a semisupervised support vector machine classification technique that exploits both labeled and unlabeled points for addressing the problem of pixel classification of remote sensing images. The proposed method is based on the transductive inference and in particular transductive SVM (TSVM). Transductive SVM progressively searches a reliable separating hyperplane in the high dimensional space through iterative method exploiting both labeled and unlabeled samples. In particular, a thresholding strategy and similarity in classification between successive transductive sets are exploited to select the reliable samples from the unlabeled set. The proposed technique is applied on two labeled datasets and one large unlabeled image dataset: IRS image of Mumbai and compared with the standard SVM and progressive TSVM (PTSVM). Experimental results confirm that employing this learning scheme removes unnecessary points to a great extent from the unlabeled set and increases the accuracy level on the other hand. Comparison is made in terms of accuracy for the numeric datasets and quantitative cluster validity indices as well as classified image quality for the image dataset.