Optimal feature selection for SVM based weed classification via visual analysis

Weed classification is a serious issue in the agricultural research. Weed classification is a necessity in identifying weed species for control. Many classification techniques have been used to identify weed based on images, however, most of the techniques only measure the percentages of accuracy but the detailed of classifier parameter are not analyzed and discussed. Therefore, in this work, feature vectors of weed images extracted using Gabor Wavelet and Fast Fourier Transform (FFT) were employed in analyzing weed pattern based on images using Support Vector Machines (SVM). The decision boundaries of the categorized extracted feature vectors are illustrated and optimal feature vectors are identified. Results are discussed and displayed with illustrations to prove the SVM classifier performance.