What is machine vision positioning? How is it different from machine vision inspection?

Visual positioning projects usually combine robot technology and axis kinematics control, and often use affine transformation, geometry, hand-eye calibration and other algorithms. At the level of mathematical principles, we should be familiar with the commonly used matrix transformation formula and geometric plane formula. In pursuit of high-precision positioning effect, the accuracy of positioning and grabbing is usually 0.01mm. Application scenarios include 2D positioning and 3D disorderly positioning and grabbing, which require a good understanding of automation equipment and robots. Machine vision detection usually refers to object detection and defect detection. In industry, it is necessary to process the image obtained by CCD sensor to find some defects. At the algorithm level, we need to master Blob analysis, preprocessing algorithm, edge extraction and so on. , and emphasizes the image processing itself. In the direction of computer vision, vision detection and target detection, convolutional neural networks are usually used to detect and classify targets, such as face recognition and automatic driving. To sum up, machine vision positioning is more inclined to the combination of visual algorithm and automation, and visual inspection pays more attention to the image algorithm itself.