Visual Dimension Inspection Project

2025-10-17

Application Overview

Visual size detection is an advanced measurement technology that utilizes high-resolution industrial cameras, precise optical lenses, and dedicated image processing software to quickly obtain the images of workpieces in a non-contact manner and precisely calculate their geometric dimensions. Visual size detection mainly measures the sizes of silicon wafers and chips, as well as the precision measurement of lead frame. The accuracy is up to 0.01mm. Through high-resolution cameras for accurate measurement, it improves product quality, reduces costs, and enhances production efficiency.

Project outcome

1) The detection speed has been increased from 10-15 seconds per item by manual operation to 2-3 seconds per item, with an efficiency increase of over 400%. This perfectly matches the production line's rhythm at high speed. The measurement accuracy is stable within ±0.05mm, far exceeding the level of manual detection (±0.1mm).

2) This workstation has achieved fully automated operation, directly saving the labor cost of 2-3 full-time quality inspectors, with a short investment return period.

3) The system automatically records the detection data, images and results of each product, and generates statistical reports. It provides solid data support for process improvement and quality analysis, achieving the transformation from "post-event inspection" to "process control".


Key technologies

1) By using high-resolution, low-distortion industrial cameras and telecentric lenses, ensure that the captured images of the remote control have clear edges and no perspective errors, laying the foundation for precise measurement.

2) According to the surface material of the remote control (such as glossy or matte) and the features to be measured (such as the edge of the button, the socket), design customized backlight or forward lighting systems. For example, using backlighting can generate high-contrast outlines, facilitating the measurement of the shape dimensions; using low-angle circular light can highlight the edges and characters of the buttons.

3) Design precise fixtures to fix the remote control, ensuring that the position and posture of the product are highly consistent each time it is photographed, eliminating measurement errors caused by positional deviations. Integrate photoelectric sensors, when the remote control reaches the detection station, automatically trigger the camera to take a photo, achieving synchronization with the production line.

4) Conduct high-precision pixel equivalent calibration, converting the pixel distances in the image to actual physical dimensions. Positioning and ROI extraction: Firstly, through algorithms such as template matching or Blob analysis, quickly and accurately locate the overall position of the remote control, and then automatically define each measured area according to the predefined detection items.

5) Use sub-pixel level edge extraction algorithms to precisely locate the contours of the remote control, the edges of the buttons, etc., increasing the measurement accuracy to the micrometer level, far exceeding the human eye's limit. Based on the extracted edge points, the algorithm automatically calculates the geometric relationships between points, lines, circles, etc., and obtains key dimensions such as length, width, diameter, center distance, and angle.


Project Background

In the consumer electronics manufacturing industry, remote controls are crucial components for many products (such as televisions, set-top boxes, air conditioners). The accuracy of the appearance size, button positions, interface sizes, etc. directly affects the user experience and assembly efficiency. The traditional manual caliper sampling method for measuring the size of the remote control housing cannot keep up with the fast production pace and can only be used for low-frequency sampling. Human judgment is subject to subjective errors, and deviations of micrometer levels are difficult to detect. The measurement data is difficult to be recorded and traced in real time, and when batch quality problems occur, it is impossible to precisely locate the problem. The requirements for the inspectors are high, which can easily lead to visual fatigue and increase the risk of misjudgment. 

To solve these problems and achieve 100% online full inspection of the production process and quality data management, we implemented the visual size inspection project for remote controls.