Abstract
Visual dimension measurement is commonly required in industrial detection. However, monocular vision measurements have a low accuracy with large field of view (FOV), while multi-vision measurement involves multi-source information fusion and incurs a large computation overhead. Hence, we propose a planar dimension measurement optimization method with metric information compensation (PMOM). This method constructs a linear representation model of pixel metric information using images at different heights along the camera’s optical axis; designs a multi-resolution multi-dimensional key point mapping mechanism (KPM) to realize the projection from a coarse resolution image point to a fine resolution image block and the secondary optimization of measurement results. By utilizing pixel-level and subpixel-level object extraction algorithms for both coarse resolution and fine resolution image blocks, the algorithm’s complexity and data processing requirements are notably reduced. The paper includes a theoretical proof of the method’s effectiveness, shown by analyzing the measurement errors of different resolution images. Comparative experiments on the measurement of spacing between planar patterns and workpiece machining accuracy validate the method meeting the requirements of industrial high-precision with a large FOV and low computational overhead.
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call