A mature citrus fruit image recognition method

Abstract

The invention discloses a method for identifying citrus fruit images in a mature period. The method comprises the following steps: collecting and cutting the images; converting the cut image into a black-white binary image, and removing black holes and carrying out black processing on a white pseudo target; Unipixelating the detected fruit contour width, and reading the fruit contour coordinates of the mark area to obtain a fruit contour coordinate matrix; forming a citrus fruit contour coordinate set arranged in a clockwise direction; finding out intersection points of different fruit contours to form a citrus fruit contour intersection point set; selecting an unshielded citrus fruit contour in each communication area and an unshielded elliptical arc section in the partially shielded citrus fruit contour; performing curve fitting: abandoning a pseudo-fruit target and removing an overfitting curve; calculating curve parameters. The method is insensitive to the angle of image acquisition, ensures the extraction accuracy of the contour of the shielded fruit area, meets the requirement of simultaneously detecting the round fruit and the oval fruit when the contour of the fruit is restored, and can simultaneously detect the partially shielded and unshielded fruit.

Description

Technical field [0001] The invention particularly relates to an image recognition method for citrus fruits in the mature period. Background technique [0002] Real-time online recognition of fruits on trees is one of the core technologies of fruit harvesting robots, and it is also a basic problem of fruit information management. The rapid recognition of occluded fruits is the first difficulty to overcome among these problems. [0003] The machine recognition of fruits on trees mainly includes three steps: extraction of unobstructed fruit area contours, restoration of fruit contours, and calculation of fruit position and size parameters. In the previous method, the contour extraction is greatly affected by the lighting conditions and the image acquisition angle. At the same time, there is the limitation of a single property curve when the fruit contour is restored. Summary of the invention [0004] The purpose of the present invention is to provide an image recognition method for ma...

AI Technical Summary

Problems solved by technology

[0003] Machine recognition of fruits on trees mainly includes three steps: extraction of unoccluded fruit area contour segments, restoration of fruit contours, and calculation of fruit position and size parameters. The previous methods are greatly affected by lighting conditions and image acquisition angles in the contour extraction process. , at the same time, there is the limitation of single property curve when fruit outline is restored

Images