Multi-view recognition of fruit packing boxes based on features clustering angle

In order to realize the intelligent mechanization of the last process of the fruit industry chains, the identification of fruit packing boxes is researched. A multi-view database is established to describe the omnidirectional attitudes of the fruit packing boxes. In order to reduce the data redundancy caused by multi-view acquisition, a new binary multi-view kernel principal component analysis network (BMKPCANet) is built, and a multi-view recognition method of fruit packing boxes is proposed based on the BMKPCANet and support vector machine (SVM). The experimental results show that the recognition accuracy of proposed BMKPCANet is 12.82% higher than PCANet and 3.51% higher than KPCANet on average. The time consumption of proposed BMKPCANet is 7.74% lower than PCANet and 29.01% lower than KPCANet on average. This work has laid a theoretical foundation for multi-view recognition of 3D objects and has a good practical application value. Copyright © by HIGH TECHNOLOGY LETTERS PRESS.