A novel semisupervised SVM for pixel classification of remote sensing imagery

摘要: This article introduces a semisupervised support vector machine classification technique that exploits both labeled and unlabeled points for addressing the problem of pixel remote sensing images. The proposed is based on applying margin maximization principle to patterns. Semisupervised SVM progressively searches reliable discriminant hyperplane in high dimensional space through iterative method exploiting samples. In particular, dynamic thresholding successive filtering set are exploited find separating hyperplane. first demonstrated six datasets described terms feature vectors then identifying different land cover regions imagery compared with standard SVM. Experimental results confirm employing this learning scheme removes unnecessary great extent from increases accuracy level other hand. Comparison made accuracy, ROC, AUC F-measure data quantitative cluster validity indices as well classified image quality data.