Plant Automatic Identification Based On Computer Vision

With the computer processing technology using in plant identification,comparing with the traditional way of collecting the data manually.It has improved efficiency effectively.this paper introduces a method of recognizing plant leaves based on shape features.Firstly we collect leaves and scan them to files with 8bits 256 colors and BMP format by scanner.And then we apply image pre-processing to get good threshold image including filtering,image enhancement,edge detection and threshold segmentation.Extract eight relative feature parameters by screening,selecting.We apply BP neural network to classify existing leaves,design neural network structure,input,and output,choose the number of hidden nodes,adjust learning efficiency,momentum factor,weight, thresholds in VC++.Succeed in recognizing leaves in VC++.The average correct recognition rate is up to 87.5 percent.