Object Detection Using Color Segmentation, Radial Symmetry Detector, and Hough Method

Currently, unmanned control systems for various vehicles, including cars, are widely used. Driving an unmanned vehicle involves solving tasks related to recognizing road objects. An improved algorithm for solving the problem of detection and recognition of round traffic light signals is proposed. To solve this problem, we use: fast radial symmetry transformation, Hough transform, and color segmentation. A special feature of the algorithm is that the areas where the light signals are located they are preliminarily determined based on the color attribute, followed by specifying the shape and position of objects. At the final stage, the solution is refined using the Hough method. The proposed algorithm has shown operability, it can be expanded in terms of the types of recognized signals and applied as part of unmanned vehicle control systems, as well as part of driver assistance systems for solving problems related to the prevention of dangerous and emergency situations.