An image processing method for green apple lesion detection in natural environment based on GA-BPNN and SVM

Abstract

The technologies of information processing, especially the image processing, are important technical means to realize intelligent management and control of modern orchard. In order to realize the intelligentization of orchard production, the orchard image database, especially the lesion image database needs to be set up. Because of the uneven illumination and the complex background under the natural light condition, traditional image segmentation methods cannot solve the adaptive threshold issue during the lesion image processing of green apples. This paper proposes an image processing method based on a BP neural network updated by genetic algorithm (GA-BPNN) and support vector machine (SVM) to realize lesion image processing of green apples in orchard. With this method, apple images can be processed in batch and the lesion image database can be consummated automatically. Furthermore, the recognition of the diseased apple is realized. The experimental results show that the proposed method can obtain green apple segmentation and lesion detection with good efficiency and robustness in complex natural orchard environment.