Method for detecting mobile phone case profile degree defects

Abstract

The invention discloses a method for detecting the mobile phone case profile degree defects. The method comprises the following steps of putting an intelligent mobile phone case (5) to be tested onto a test platform in the front center position of a test system structure element (4); a first CMOS image pickup module (2) and a second CMOS image pickup module (3) are arranged at the two ends of the test system structure element (4). The concrete method comprises the following steps of 1, putting the mobile phone case onto the test platform; using a semiconductor line structure optical laser with the blue wavelength being 405nm for illuminating the mobile phone structure element; respectively collecting laser line images by the first and second CMOS image pickup modules; obtaining the upper surface 3D point cloud data; selecting the camera optic axis and laser line optical plane included angle to be 45 degrees; 2, setting the sensor sampling line number being n pixels; determining the sampling pixel line numbers and the like of the sensor according to the depth measurement range.

Description

technical field [0001] The invention belongs to the field of manufacturing and relates to a method for detecting contour defect of a mobile phone shell. Background technique [0002] With the development of mobile terminals (mobile phones), the quality requirements for mobile phones are getting higher and higher. In the past, most manufacturers used visual observation to detect the appearance of the bottom of mobile phones, which is very subjective and difficult to stabilize the quality. In particular, the human eye is not sensitive to surface contour deformation, and it is difficult to detect small-sized bumps and convex points, resulting in more missed inspections. , but consumers can find bump defects through finger touch. A small number of companies in the industry use machine vision to detect the appearance of mobile phones, such as Huawei Technologies Co., Ltd., Samsung, etc., and have also deployed some test equipment to the mobile phone production line. The mainstre...

AI Technical Summary

Problems solved by technology

However, in actual production, since the occurrence of defects is a low-probability event, the number of defect samples is often small, and the training takes a long time. It often takes hours or even days for training, so it takes a long time for new models to be introduced into the detection system.

[0004] 2. The current laser profilometer can output surface 3D point cloud. The main profilometer manufacturers include Keynece, Sick, etc., but their system design indicators tend to expand the range of depth and distance. The width of one measurement is narrow and requires multiple times. Scan the phone case to meet the requirements

Directly used for surface detection has poor robustness, poor anti-interference, and high false detection rate. Directly using point cloud data as a reference model for detection is easily affected by the vibration of the motion platform, and calibration errors will also have a great impact

[0006] 4. There are many features that reflect product defects, and manual judgment has great ambiguity. Conventional linear machine judgment methods are prone to missed judgments and misjudgments

There is also some uncertainty in using machine learning at all

Images