Abstract: A number of color recognition algorithms are known - from using classic HSL space to modern AI algorithms. However, when analyzing real-life industrial, agricultura,l or medical images (for example images of fruits, flowers – to be sorted by color, human skin – to be analyzed for the presence of suspicious colors) - camera noise makes color recognition non-reliable, especially in dark regions. In this paper reliability of a number of well-known color recognition algorithms was compared. In order to lessen the amount of false recognitions, some modifications to algorithms in test were added. Provided comparisons of the algorithms in test were executed by using real-life images.
Keywords: image processing, color recognition, hue, scalar product, normalized correlation
Cite this article: Samuel Kosolapov. COMPARISON OF ROBUST COLOR RECOGNITION ALGORITHMS. Journal of International Scientific Publications: Materials, Methods & Technologies 15, 274-283 (2021). https://www.scientific-publications.net/en/article/1002218/
