A quality detection method and detection system for a finished product of a heavy-duty automobile connecting rod

Abstract

The invention discloses a detection method for the quality of a heavy-duty car connecting rod finished product. The detection method for the quality of the heavy-duty car connecting rod finished product comprises the following steps: a company logo image on a heavy-duty car connecting rod is collected in advance, and the information of the company logo image is extracted and stored as prior information; a car connecting rod image containing a company logo and a product serial number on the heavy-duty car connecting rod to be detected is collected in real time; the speeded-up robust feature (SURF) eigenvector of the car connecting rod image is calculated, and an interest area containing the product serial number is determined; the interest area is divided into a plurality of subordinate areas; the integral image of each subordinate area is calculated, the SURF eigenvector of each integral image is extracted, and each subordinate area obtains a pair of SURF eigenvectors; the Euclidean distance between each pair of SURF eigenvectors is calculated to judge whether the heavy-duty car connecting rod to be detected is qualified. The quality of the heavy-duty car connecting rod finished product is detected accurately in real time, the detection efficiency and detection accuracy are high, and labor intensity for detectors is largely relieved at the same time.

Description

Technical field [0001] The invention belongs to the technical field of product quality detection, and specifically relates to a detection method and a detection system for characters on the surface of a heavy-duty vehicle connecting rod. Background technique [0002] The connecting rod is a key component of the engine, and its product quality is very important. A single heavy-duty vehicle connecting rod is assembled from a connecting rod body and a connecting rod cover. Each connecting rod body has a unique corresponding connecting rod cover. If you assemble different pairs of connecting rod bodies and connecting rod covers into At the same time, defective products will be produced. [0003] The product quality inspection process of heavy-duty truck connecting rods needs to be able to quickly and accurately detect the registration of connecting rod bodies and connecting rod covers. Real-time online inspection methods are very important for timely and efficient quality of connectin...

AI Technical Summary

Problems solved by technology

However, due to factors such as product processing technology, the quality inspection of finished connecting rods has become a technical problem in automatic product quality inspection

[0004] Corresponding product serial numbers on the connecting rod body and the connecting rod cover Under the existing process conditions and technical background, the matching quality inspection of the connecting rod generally adopts the offline manual naked eye detection method, and the operator visually inspects the connecting rod body and the connecting rod cover. Whether the corresponding product serial number is the same to determine whether the connecting rod is qualified. Although the detection requirements are met to a certain extent, the product serial number on the connecting rod body and the connecting rod cover is long and the numbers and letters are mixed. There are two problems: one is that people are prone to fatigue and mistakes in the long-term monotonous and boring detection process; the other is that the degree of automation of naked eye detection is low, and the production efficiency is low, which cannot meet the actual needs of enterprises for fast and accurate detection.

[0005] Therefore, the domestic heavy-duty automobile connecting rod manufacturers are currently unable to realize the automatic quality inspection of the finished connecting rod. The simple manual inspection is time-consuming and laborious with low accuracy and poor reliability.

[0006] Bay H. and Tuvtellars T. disclosed a sparse local based on SURF (Speeded Up Robust Features) in an article titled SURF:speeded up robust features (Proceedings of the European Conference on Computer Vision, 2006:404-417). Invariant features are used to describe the shape of images. This method is fast and accurate for image target detection, but it is only successfully applied to face detection and tracking.

Images