Edge-detection method for low-quality color pictures in lossy compression image file formats

The objective of this article is to propose a robust edge detection algorithm appropriate for the processing of color images originating from image acquisition systems with unknown parameters, with the extreme case of low-quality color pictures in lossy compression image file formats taken from the internet, and to assess the performance of the proposed solution in comparison with the Canny edge detection algorithm, using a proof-of concept software developed in Python. The output of the software is intended to feed unassisted AI processes related to artificial vision; thus, it also involves a method of establishing a hierarchy in the detected edges and an identification strategy for reestablishing the continuity of the curves that can be assimilated with mathematical primitives and to de-fractalize the noise-induced features.