Rose Flower Detection and Feature Extraction Based on Machine Vision
Roses in the planting environment are closely distributed and shielded from each other.In order to accurately detect and extract the characteristics of roses,the colour and shape of roses were recognized and processed based on machine vision.Firstly,bilateral filtering was selected to denoise the rose image,and then the colour of the rose was extracted by using the hexagonal cone colour model(HSV),and a scrollbar function was created to segment the threshold of each component of the hexagonal cone colour model to determine the optimal threshold.Finally,the contour of the rose was extracted by means of morphological operation,area threshold,hole query and filling.The fitting method of the shape of the inner circle of the rose was proposed,and the center and radius of the fitting inner circle were used as the image features of the rose.The results showed that the rose colour threshold could effectively remove the images of the rose branches and leaves,and the shape fitting algorithm could effectively extract the shape features of the rose and erase the rose bud.By using this method,the recognition rate of single rose was 98.17%,that of overlapping roses with 3 or less roses was 92.67%,that of overlapping roses with 4 or more roses was 74.07%,and that of roses blocked by branches and leaves was 83.03%.This set of machine algorithm could effectively recognize and extract the characteristic values of roses in a complex planting environment,which provided important technical support for rose-picking robot.
国家科技期刊平台
NSTL
万方数据
维普
