Image recognition of camellia fruit based on preference for aiNET multi-features integration

To promote the recognition rates of machine-vision system in camellia fruit picking robot, the paper proposed the method of multi-feature integration using preference aiNet as the recognition algorithm. Based on the image procession of camellia fruit, the color feature, the morphology feature and the texture feature of the object region were clustered and the multi-features of camellia fruits were taken as the preference antibody, so the parameters of multi-features made effective integration in the preference aiNet. The simulation proved the accuracy of multi-feature integration reached 90.15% in the sunny day and 93.90% in the cloudy day, and the time complexity was nearly the same. This research has a certain meaning in application of forestry picking robot and provided the reference for further research of the camellia fruit picking robot.