Injection-molded product defect detection and recognition method

Abstract

The invention discloses an injection-molded product defect detection and recognition method, which comprises the following steps: 1) images of a normal injection-molded product and a known defect-existing injection-molded product are acquired, and the images are classified and samples are generated; 2) a multi-layer convolutional neural network model is built; 3) the samples in the first step are used for training the convolutional neural network model built in the second step; and 4) the acquired actually-measured injection-molded product images are preprocessed and inputted to the convolutional neural network model after training completion in the third step, the convolutional neural network model carries out classification and recognition on the actually-measured injection-molded product images so as to judge whether injection molding defects exist in the injection-molded product, and if the defects exist, the defect types are classified. The built convolutional neural network model of the invention adopts the same feature selection and extraction mode for various injection-molded defects, is strong in adaptability, and can be used for detection and recognition of various types of defects of the injection-molded product.

Description

technical field [0001] The invention belongs to the technical field of defect detection and identification, and more specifically relates to a defect detection and identification method for injection molded products. Background technique [0002] During the injection molding process, due to factors such as changes in the physical parameters of raw materials, unreasonable process parameters, and insufficient performance of the injection molding machine, injection molded products will have defects such as short shots, flashes, cracks, warping, bubbles, and weld lines. These defects not only affect the appearance of injection molded products, but also seriously affect their performance. However, the current defect detection and identification of injection molded products mostly rely on manual offline sampling detection, the degree of automation is not high, the detection efficiency is low, and it is easily affected by the subjective will of the inspectors, and the accuracy rate...

AI Technical Summary

Problems solved by technology

[0004] In view of the above defects or improvement needs of the prior art, the present invention provides a defect detection and recognition method for injection molded products, which automatically performs feature selection, extraction and learning based on the convolutional neural network model, without manual selection and extraction of features. Strong adaptability, suitable for detection and identification of various types of defects in injection molded products

Images