Green apple recognition method based on the combination of texture and shape features

Automated harvesting requires accurate recognition of fruit in a tree canopy in uncontrolled environments. However, occlusion, variable illumination, variable appearance and texture make this task a complex challenge. Therefore, an accurate recognition algorithm needs to be studied which involves the detection of green apples within scenes of green leaves, shadow patterns, branches and other objects found in natural tree canopies. In this paper, the method is proposed which combines texture features, shape features and color features to solve the problem of segmenting the target and background of apple picking robot in complex background. The gray-scale difference statistical method is utilized to get the texture feature vector of the image. According to the texture feature vector, the support vector machine (SVM) is used to segment the image preliminar, and then the shape and color features are combined to achieve precise segmentation. Experiments show that the algorithm of this paper is superior to other algorithms in recognition rate and running speed. In addition, it has better segmentation effect for fruit with slight background occlusion.