The working principle of the machine vision system can be simply understood as "the machine sees with its eyes and analyzes with its brain". It mainly acquires the target image through image acquisition devices (such as industrial cameras), and then analyzes and judges the image through image processing algorithms to achieve the purposes of detection, identification, positioning, measurement, etc. The following is the basic workflow of the machine vision system:
1. Components
- The light source provides stable and appropriate lighting to make the features of the target object clearer.
- The industrial camera (image acquisition device) captures the image and converts the light signal into an electrical signal. Common ones include black and white cameras, color cameras, line array cameras, and area array cameras.
- The lens determines the clarity, magnification, and field of view of the image, and plays a role in focusing and imaging.
- Frame Grabber / Interface Standards (such as USB3.0, GigE, Camera Link, etc.) are used to transmit the images captured by the camera to the processing system.
- The image processing unit (usually an industrial computer or embedded computing platform) runs the machine vision algorithm to analyze, judge, and make decisions on the image.
- The software system includes image processing software, algorithm modules, user interfaces, etc., which are used to complete specific application tasks.
- The execution unit (such as PLC, robot) performs specific actions according to the processing results, such as sorting, rejecting defective products, positioning and assembly, etc.
2. Workflow
- Image acquisition: The camera shoots the target object at a specific moment, forms an image through the lens and collects the image.
- Image preprocessing: Such as denoising, contrast enhancement, edge extraction, grayscale transformation, etc., to improve image quality.
- Feature extraction and analysis: Identify the edge, shape, color, barcode, character and other information of the target area.
- Judgment and output results: Make judgments according to the set standards (such as whether the size is qualified, whether the position is correct), and transmit the results to the control system.
- Feedback control execution: The controller drives the mechanical action (such as rejection, grasping, stopping, etc.) according to the visual results.
3. Application field
- Quality inspection: Detect product appearance defects, dimensional errors, etc.
- Positioning guidance: Provide accurate grasping or assembly positions for industrial robots.
- Recognition and classification: barcode recognition, character recognition (OCR), color recognition, etc.
- Measurement: high-precision measurement of geometric parameters such as size, angle, flatness, etc.
