Color segmentation and color histograms

Color segmentation 20210709195948 is a common technique used on robotics 20210602191112 due to being a fast technique that provides good results. The image is segmented into a binary image where regions with the desired color (values are between certain thresholds) are marked. We can then try to measure, for example, the biggest region or the center of the defined region.

This technique is, however, dependent on the environment lighting leves, as rooms with more or less light intensity require different thresholds.

Another technique are color histograms. Color histograms measure each of the pixels and categorizes them into “buckets,” each one representing a certain color. We can use this histograms to detect if there is a certain object of interest, as we can comprare different histograms, one for the current image and one for a certain objective with a certain histogram.

Notes References

20210602191112 Cameras and Computer vision in Robotics - Basics

20210709195948 Computer Vision - Segmentation

20210709195051 INDEX - Computer Vision

References

(Murphy 2000)

Murphy, Robin. 2000. Introduction to AI Robotics. Intelligent Robotics and Autonomous Agents. Cambridge, Mass: MIT Press.