Identification of Sugarcane Bud Based on Image Processing and BP Neural Network

Sugarcane bud is an important part of sugarcane seeds. In order to analyze its morphological characteristics and improve the automatic recognition rate of sugarcane buds, according to the characteristics of sugarcane seeds images, a method of extracting sugarcane using bilateral filtering combined with seed region growth (RSG) was verified.The method of planting cane buds: first use a bilateral filter to smooth the cane buds while ensuring the edges, and then use RSG to segment the cane buds.Choose Guitang No. 44, which is commonly used in Guangxi, as the test object. The area and circumference of the cane bud area extracted by this method are counted, and linear regression analysis is performed with the manually measured area and circumference. The mean values of the correlation coefficient R2 are respectively.Reached 0.9579, 0.9885.By extracting 17 parameters of the color feature and shape feature of the sugarcane bud image as the input of the BP neural network, the sugarcane bud region recognition is realized.The experimental results show that the average recognition rate of the sugarcane buds of the sugarcane test subjects is 96.4%, which has achieved good recognition results and has certain practical value.