Pixel-based Scene Analysis in Robot Vision

Abstract

This study is in the area of image processing, particularly in the field of color detection. In this work, the researchers propound and implement a novel algorithm for color detection. The researchers explain how the algorithm works in any type of scene and finds the locality of any desirable color. The implementation of the algorithm is in a simple robotcamera that is able to move toward the desirable color vertically. As the algorithm of the system plays with images of a video being captured on-time, the robot works real-time. The algorithm of the robot takes some advantage of partitioning of the frames and then considering all pixels located at each partition to distinguish the locality of the desirable color. The algorithm does not contain any time-consuming method such as transformation related to computer vision. This is why; the robot works fast. This study also explains all computations regarding movement of the robot. The researchers tested the robot regarding right movement and left movement separately and found out different accuracy; 79.5% accuracy for right movement and 89.7% accuracy for left movement.