2D Image Processing for Randomly Stacked Workpiece Position Estimation

The patent uses two-dimensional images captured by cameras as copies or representations of the three-dimensional workpiece arrangement. Instead of directly measuring workpieces with expensive three-dimensional measurement machines, the system captures images and processes them to extract position and orientation information, effectively using a lower-cost proxy (2D image) to achieve the same measurement function.

The patent replaces the mechanical three-dimensional measurement system with an optical system (camera) combined with image processing algorithms. The mechanical measurement process is substituted by capturing light reflected from workpieces and computationally deriving spatial information through feature extraction and matching algorithms.

The patent makes the camera system universal by enabling it to perform multiple functions: capturing images for position detection, extracting feature information, and processing spatial data. A single camera system can handle various workpiece configurations and measurement tasks, eliminating the need for multiple specialized three-dimensional measurement machines.

The patent segments the measurement process into distinct computational steps: capturing images, extracting features from images, matching features between images, and calculating position and orientation. This segmentation allows the system to handle complex measurement tasks through a series of simpler processing stages rather than requiring complex hardware systems.