Aircraft bearing surface rivet detection method based on U-net network

Abstract

The invention discloses an aircraft bearing surface rivet detection method based on a U-net network, particularly relates to the field of machine vision detection, and aims to solve the problems thatrivet assembly detection of an aircraft bearing in the prior art can only be achieved through manual detection, and the detection efficiency is low. The method comprises the following steps: collecting an image of a rivet with standard assembly as a standard template of a rivet image, and collecting an image of a rivet on the surface of an aircraft bearing as a training sample; constructing a U-net network; inputting a training sample into the U-net network for training to obtain a trained U-net model; inputting a to-be-detected image, and carrying out the feature recognition of the to-be-detected image through the trained U-net network; performing circle identification on the segmentation result by using Hough transform, and outputting the radius of the identified circle; performing rivetinstallation judgment; according to the invention, manpower and material resources are saved, the detection speed is improved, and the detection process is more automatic and intelligent.

Description

technical field [0001] The invention relates to the field of machine vision detection, in particular to a method for detecting rivets on the surface of aviation bearings based on a U-net network. Background technique [0002] Aviation bearings are very important basic parts in aviation equipment, and their performance plays a decisive role in the operation of many mechanical components. During the assembly process, sometimes certain assembly errors occur due to various reasons, resulting in the lack of rivets on the bearing surface, which in turn affects the performance of the entire bearing. In some cases, this lack may cause more serious consequences, so it is particularly important to perform rivet inspection on assembled bearings. At present, many rivet inspections still rely on manual means, but the inspection speed is slow, and it is prone to missed inspections, which sometimes causes serious safety hazards. In order to seek a more automated detection method, a detec...

AI Technical Summary

Problems solved by technology

For rivet detection based on machine vision, if the circular feature in the image is not obvious, or the edge of the rivet on the bearing surface is not clear, it will affect the detection accuracy

And due to the collection angle problem, the riveting interface of some nail-less targets is also similar to a circle, and the size of the circle needs to be determined, so the detection has higher requirements for the accuracy of feature extraction

Images