Abstract: Visual-Label is a novel concept of Intelligent Vision Internet of Things, which is widely used in areas such as face recognition as well as license plate recognition. However, application of plant's visual-label is hot and difficult in intelligence research area. So, a method for extracting feature vectors of the neighborhood of corner points, combined with value of local binary patterns used to establish the visual-label for plant leaves is proposed. Firstly, we detect the corners of different states of plant leaves, extract the value of local binary patterns as the center of the corners and calculate its neighborhood gradient vector, and then assign gradient direction and right weight to get an eighteen-dimensional feature vector. Finally, we calculate the angle of feature points feature vectors between the different leaves which have the equal values of local binary patterns. Matching plant leaves by the value of angle, which judges the state of plant leaves. This feature works well for the plant visual-label and is of good robustness for the blight caused by plant diseases and insect pests and the rotation of the blade.

Key words: visual-label, corner detection, Local Binary Patterns, gradient vector

 visual-label, corner detection, Local Binary Patterns, gradient vector