Red date long diameter detection method based on OpenCV

Abstract

The invention discloses a red date long diameter detection method based on OpenCV, and the method can effectively obtain the size of the long diameter of a red date through the combination and improvement of an OpenCV open source library minimum enclosing rectangle algorithm, and in combination with a Matplotlib drawing library. Firstly, red dates are placed on a conveying belt, and images of the red dates are collected through an industrial camera installed above the conveying belt; graying processing is carried out on the collected image, mean filtering and binarization processing are carried out, coordinates of a central point and four vertexes of a minimum enclosing rectangle of each contour are solved, each side of the rectangles is drawn by using a Matplotlib drawing library, and the maximum pixel value in each side of the rectangles is solved, namely the size of the long-diameter pixel size value of the red dates; and finally, determining the size of the long diameter of the red date according to the relation between the pixel size value of the long diameter of the red date linearly fitted by the least square method and the actually measured long diameter value. The detection efficiency and precision can be greatly improved, and then accurate grading of the red dates is achieved.

Description

technical field [0001] The invention relates to a method for sorting and detecting fruits, in particular to a method for detecting the long diameter of red dates based on OpenCV. Background technique [0002] At present, most of the sorting red dates on the market pass through meshed slopes or drum sieves. This sorting method can only sort red dates of different sizes. The sorting standard is single, and the efficiency is not high. Red dates are accurately graded during the sorting process. Patent application CN106423895A discloses a jujube classification method. After the jujube image collection is completed, the construction of a jujube mask image includes using a threshold to binarize the image, and then performing morphological opening and filling operations to remove the noise interference in the target. The illumination-reflection model uses low-pass filtering to obtain the brightness of the R component image and uses the brightness component to correct the brightness...

AI Technical Summary

Problems solved by technology

[0002] Most of the sorting jujubes currently on the market pass through mesh slopes or drum sieves. This sorting method can only sort jujubes of different sizes. The sorting standard is single, and the efficiency is not high, wasting time, and it is impossible Precise grading of red dates during sorting

Patent application CN106423895A discloses a jujube grading method. After the jujube image is collected, constructing the jujube mask image includes binarizing the image using a threshold, and then performing morphological opening and filling operations to remove the noise interference in the target. The illumination-reflection model uses low-pass filtering to obtain the brightness of the R component image and uses the brightness component to correct the brightness of the component image after removing the background. The processing method is more complicated, and the detection rate of defective products is not high.

Images