Intelligent Real-Time Image Processing Technology of Badminton Robot via Machine Vision and Internet of Things

The development of precise detection and tracking methods for high-speed moving targets, coupled with accurate matching under varying lighting conditions, represents a formidable challenge in the realm of badminton robot creation. To tackle these intricate issues, this work introduces a suite of innovative techniques within the domain of intelligent robot image processing, offering substantial impetus to the evolution of badminton robots. Primarily, the work designs a high-precision binocular stereo vision acquisition system to enhance the accurate reconstruction of the badminton shuttlecock’s trajectory in three-dimensional space, enabling real-time monitoring and in-depth analysis of shuttlecock motion. Subsequently, an inventive dynamic threshold Gaussian mixture model, underpinned by singular value decomposition, is introduced. This model dynamically adjusts the decision threshold based on background-foreground similarity while ensuring synchronized exposure between the left and right cameras and timely data retrieval. Lastly, the study addresses the impact of lighting fluctuations on target matching through a Retinex-based color histogram matching method, adapting color histogram templates according to varying lighting intensities. Experimental validation involves the utilization of an industrial digital camera with frame exposure to capture badminton images, subsequently transforming image processing into motion control to effectively realize the badminton robot’s hitting action. The experimental outcomes highlight the following achievements: 1) The designed badminton robot achieves a remarkable 100% recognition rate for the shuttlecock under normal conditions, featuring a standard deviation of
$\theta $
below 0.05°, a position standard deviation of less than 18 millimeters in the x-axis direction, and less than 8 millimeters in the y-axis direction; 2) The proposed binocular stereo vision acquisition system exhibits superior precision, profoundly supporting shuttlecock trajectory analysis; 3) The Gaussian mixture model, incorporating dynamic threshold adjustment via singular value decomposition, excels in detecting delicate and diminutive moving targets, showcasing high-level detection accuracy. These experimental findings offer novel insights into the domain of intelligent real-time image processing technology for badminton robots anchored in the IoT (Internet of Things) machine vision. The implications reverberate through the further advancement of badminton robots, charting a novel technical trajectory that amalgamates intelligent real-time image processing with the IoT. Consequently, the amalgamation bolsters hitting accuracy and motion performance, cohering with the ongoing trajectory of robotics technology development, which emphasizes the integration of intelligence and autonomy across diverse facets of robotics. This alignment propels continuous progress throughout the entire expanse of robotics technology.