Binary Color Vision for Industrial Automation.

摘要

This project addresses the questions of the near-term utilization of color by existing state-of-the-art binary vision systems. Using the experimental testbed, it was demonstrated that under suitable thresholding and filtering, the spectral contents of color scenes could be artificially controlled. Rejection of unwanted colors or background clutter could be performed. Under single threshold selection, discriminations of objects identical in shapes and sizes but different in colors could be achieved by utilizing the luminance effect of colors. Color encoding combines the description of colors and thresholding and provides a simple interface to computer processing. The introduction of color discriminations for current machine vision processing can significantly extend their range of applications. Experiments using color filters to separate the natural image into color component images show that binary image processing is a valuable next step in color region recognition. (Author)