The Mechanism of Motion Direction Detection from Gray Images

Yuxiao Hua, Riku Inoue, Cheng Tang, Yuki Todo, Zheng Tang

Abstract: The visual system plays an important role in the daily physiological activities of human beings. Because most of the information that people receive from the outside world is generated by the visual system. And the detection of motion direction is the key for researchers to simulate the human visual system by computer. However, the mechanism of motion direction detection remains mysterious. In this paper, a motion direction detection mechanism that can derive global orientation from local orientation information obtained by cells in the retina is proposed to achieve motion direction detection in humans. Assuming that there are neurons in this mechanism that respond to the local movement in each specific direction. The main idea of this mechanism is to derive the global motion direction by extracting local motion direction information of the object from these neurons. The mechanism is simulated by the dendritic neuron model, and we verified the accuracy by conducting various experiments. The results that the mechanism accurately detected the motion direction of the object in all experiments show that this mechanism can accurately detect the motion direction of objects with different shapes, orientations, positions, and sizes in a grayscale background.