HIGH-SPEED REGISTRATION AND REACTION TO OBJECT PENETRATION INTO THE RASPBERRY PI CAMERA FIELD OF VIEW

Abstract

Software and an experimental setup have been created to investigate the high-speed penetration of a small object into a specified control zone and the system's reaction, particularly by issuing a corresponding signal. A feature of the system is the developed algorithm for analyzing the minimal necessary number of pixels on the virtual perimeter of the single-board computer camera matrix. This is based on tracking changes in their intensity without processing the entire or partial volume of the matrix. This approach allows for instantaneous reaction on low-power computer platforms. The program is implemented in Python using the OpenCV and NumPy libraries. For the research, a custom-designed experimental stand based on the Raspberry Pi 4 single-board computer with a Pi V2 camera was used. The stand includes a slider for attaching and controlled movement of multi-colored object mock-ups, specifically in the form of a triangle, disk, and square, through the camera's field of view. Multi-colored sheets of paper (white, green, gray, etc.) were used as the object movement background. The movement speed of the object mock-up was calculated based on the ratios of distances from the camera lens to the background plane of the mock-up and to the movement plane of real objects (sports balls, birds, animals, etc.), as well as the speeds of these objects, and was of the order of 1–10 cm/s. The program records detailed pixel parameters (coordinates, color parameters R, G, B, brightness) to files at the moments of object detection when it penetrates the observation contour. The obtained arrays of numbers (0–255) are used in a graphical editor like Origin for visualization and analysis of penetration processes. The developed software and experimental stand achieved high-speed object detection (approximately 1 millisecond) by processing a minimal number of pixels on the virtual control perimeter (about 700 points), with a reaction signal trigger time of approximately 1.5 milliseconds (which is significantly less than the single frame duration of 33 milliseconds).