Plant Disease Recognition Using Fractional-Order Zernike Moments and SVM Classifier

Abstract

Orthogonal moments have a vital role in the field of digital image processing and analysis. These moments are derived from statistically independent orthogonal polynomials and can be continuous or discrete. Major continuous orthogonal mo- ments are Zernike, Pseudo-Zernike and Legendre moments and the primary dis- crete moments includes Krawtchouk, Tchebichef and Racah moments. In almost all researches till date, orthogonal moments are used with order having integer value, however, there is very less research on application of orthogonal moments with non-integer order which are also known as fractional-order orthogonal mo- ments. Orthogonal moments with integer-order are a special case of fractional- order orthogonal moments. Each orthogonal moment can be made of fractional- order type by replacing integer order with real order. This study proposes the use of fractional-order Zernike moments in the recognition of grape leaf diseases and comparative analysis of these moments with integer-order Zernike moments is also presented. After extensive experiments, it can be said that the proposed technique is more robust to image noise and shows higher recognition rate (95.12% at order 30) compared to conventional techniques.