Machine vision system refers to the conversion of the captured object into image signal through machine vision equipment, which is transmitted to a special image processing system. According to the information of pixel distribution, brightness, color and so on, it is transformed into digital signal. The image system carries out various operations on these signals to extract the characteristics of the target, and then controls the device action on the spot according to the result of discrimination.
Machine vision is to use machine instead of human eyes to measure and judge. In some dangerous working environment or unsatisfactory environment, machine vision is often used to replace human eyes.
一个典型的工业机器视觉系统包括：光源、镜头、CCD 相机和COMS相机、图像处理单元（或图像捕获卡）、图像处理软件、监视器、通讯 / 输入输出单元等。
表面缺陷检测设备，包括尺寸、角度、个数、合格 / 不合格、有 / 无等，实现自动识别功能。
Machine vision inspection can improve the detection accuracy and production efficiency. In the case of large number of products and low detection accuracy, the manual detection efficiency is low, and the accuracy is far from enough for the machine. Therefore, more and more machine vision inspection equipment began to replace the manual.
A typical industrial machine vision system includes light source, lens, CCD camera and COMS camera, image processing unit (or image capture card), image processing software, monitor, communication/input/output unit, etc.
The working principle of machine vision detection system is to use CCD camera to convert the detected products into image signals, and then transmit them to industrial computer image processing system. According to the information of pixel distribution, brightness, color and so on, the system transforms them into digital signals. The image processing system carries out various operations on these signals to extract the characteristics of the target, such as area, length, quantity, location, and then. According to the preset permissibility and other conditions output results, including size, angle, number, qualified/unqualified, yes/no, etc., to achieve automatic identification function.
What are the main applications of machine vision?
The application of machine vision includes detection and robot vision.
1. Detection: It can also be divided into high-precision quantitative detection and qualitative or semi-quantitative detection without gauge. High-precision detection, such as cell classification of microphotographs, size and position measurement of mechanical parts; non-high-precision detection is more prominent in industrial areas, such as product appearance inspection, parts identification and positioning on assembly line, defect detection and assembly integrity detection.
2. Robot vision: It is used to guide the operation and action of the robot in a wide range, such as picking up the workpieces from the messy workpieces sent out by the hopper and placing them on the conveyor belt or other equipment in a certain direction (i.e. hopper picking problem). For small-scale operations and actions, tactile sensing technology is also needed. In addition, there are: (1) automatic optical inspection (2) face detection (3) driverless vehicle.