Shielded vegetable and fruit harvesting method based on depth association perception algorithm

Abstract

The invention provides a shielded vegetable and fruit harvesting method based on a depth association perception algorithm. The shielded vegetable and fruit harvesting method comprises the steps of collecting an RGB channel color image, a D channel depth image and an I channel infrared image; unifying the resolution of the acquired image; removing areas except the working area of the harvesting robot in the image with the uniform resolution; identifying a mature target, and marking each pixel of the mature target in the cut image; judging whether the mature target is shielded or not, restoringthe images of the RGBD four channels of the complete mature target under the condition that the mature target is shielded, and directly obtaining the images of the RGBD four channels of the complete mature target under the condition that the mature target is not shielded; and according to the images of the four channels of the RGBD of the complete mature target, extracting the position and postureof the mature target to a harvesting robot. The device is less interfered by the environment and works more stably; the complete RGBD information of the target can be estimated; robot harvesting success rate can be improved.

Description

technical field [0001] The invention relates to the field of image processing, in particular to a method for harvesting occluded fruits and vegetables based on a depth associative perception algorithm. Background technique [0002] There is a large labor gap in the large-scale production of vegetable and fruit picking operations. The use of robots for automatic harvesting can improve production efficiency, but the technical requirements are high. At present, there are few large-scale applications of robots for precise picking of fruits and vegetables. An important limiting factor is that the robot picking requires the target to be unobstructed in the field of vision. However, in the actual agricultural environment, the environment is complex, and there are a large number of branches and leaves blocking the target, which reduces the success rate of robot picking. . On the other hand, the shape of vegetables and fruits is changeable, and it is difficult to extract the control...

AI Technical Summary

Problems solved by technology

Both of the above two schemes require the target to be unobstructed in the field of view, and all redundant branches and leaves need to be removed. It takes a long time to judge whether the fruits and vegetables are ripe. At the same time, the actual operation is seriously affected by the light, so the picking success rate is low and difficult. widely used

Images