Research On Fruit Axial Position And Orientation Based On Feature Points By Visual Measurement Method

With the continuous development of robot technology, the machine vision system has been more and more applications in automatic detection in the quality of agricultural products and fruit grade separation. At present, machine vision technology is becoming more widely used to identify the location of the work-piece profile in industrial robots applications such as welding, packaging, palletizing and assembly. However, researching on fruit'three-dimensional space position and orientation by visual method is not much, this research is benefit for automation of fruit picking, processing and other operations, it can not only improve fruit processing hygiene, also can reduce labor intensity and improve labor efficiency, it is a good choice.This topic is for the fruit on the production conveyor belt, taking grapefruit as the studying object, building binocular vision measurement system to calibrate the parameters of the system and get fruit images, then do preprocessing, feature extraction, combined with photography geometric principle to get the fruit space geometric parameters, the solving results will be proved whether meet the engineering requirements by experiment, and eventually according to the final solution of a group of fruit space geometric parameter values, gives the joint variables value when the robot to crawl fruit. The main contents are as follows:Carrying out experiments on the fruit images using various pretreatment techniques, getting suitable image pre-processing methods for the system through the analysis of comparative test results, through the feature extraction method to extract the fruit edge, calyx point, and the approximate size.According to judge the grapefruit placed according to the got images, then based on the fruit placement to select the solving method for orientation parameter. The orientation parameters will be defined for vertical placement of grapefruit; for tilted grapefruit, using a relatively simple and quick method of space plane intersection. The method was verified effectiveness by experiments.According to propose an approximate match and combined with photography geometric principles to solve its three-dimensional coordinates, the method was verified its feasibility. According to calibrating the fixed-length template, getting the conversion equivalent of the pixel length to the actual length, then solving the fruit size, which was proved effectiveness by experiments.Through the introduction of production line and the analysis of degrees of required grabbing grapefruit, determining to use six degrees of freedom robot to crawl grapefruit, analyzing the simple kinematics of the robot, according to a group known space geometric parameters of grapefruit giving the joint variables value when the robot to crawl fruit.