A texture feature extraction of crop field images using GLCM approach

Abstract

To capture visual content of images for retrieval, feature extraction is one of the method. In this paper feature extraction is done using GLCM (Gray Level Co-occurrence Matrix). In this work 6 varieties of crop images are considered namely paddy, maize, cotton, groundnut, sugarcane and sunflower. There are many second order statistical texture features extracted using GLCM namely autocorrelation, entropy, cluster prominence etc. The four features namely autocorrelation, sum of squares of variance, sum of variance and sum of average are found to be predominant features for the present study. Considering texture as a feature, the average accuracy of 63.75% is obtained. The results show that these texture features are efficient and can be used for real time pattern recognition.

Keywords: Field images, GLCM, Features, Pattern recognition