Apple leaf disease recognition based on feature fusion and local discriminant projection

As for the complexity of plant disease recognition by the disease leaf image，a plant disease recognition method was proposed based on feature fusion and local discriminant projection. First，based on the center symmetric local binary pattern（CS-LBP），an adaptive CS-LBP algorithm was proposed. The spot images were segmented by ACS-LBP. The texture，shape and color features were extracted from each spot image and fused，and then were reduced based on local discriminant projection（LDP）. Finally，the diseases were recognized by support vector machine（SVM）. The experiment results on a database of apple disease leaf images showed that the proposed method was effective for apple leaf disease recognition. The average recognition rate was more than 95%.