MODIFICATION OF TEXTURE DETECTION METHOD USING FEATURE AGREGATING

Abstract

The importance of texture area detection in images is motivated. A modification of the method of texture area detection using SVM-classifier for images corrupted by rather intensive additive white Gaussian noise by aggregation of input parameters is proposed. A modified procedure of classifier training is described. Criteria of processing efficiency used in analysis are given. For test data, the effectiveness of the developed approach and improvement of classifier performance are demonstrated

Keywords

• текстура
• обнаружение
• изображение
• помехи
• обучение