Cloud Computing Image Processing Application in Athlete Training High-Resolution Image Detection

• Published in: Computational intelligence and neuroscience Vol. 2022; pp. 1 - 12
• Main Author: Li, Hongtao
• Format: Journal Article
• Language: English
• Published: New York Hindawi 03.10.2022; John Wiley & Sons, Inc
• Subjects: Algorithms; Applications programs; Artificial intelligence; Athletes; Augmented reality; Biometry; Cell phones; Cloud computing; Communication; Computer programs; Decision making; Electronic devices; Energy consumption; Environmental factors; Equipment and supplies; Euclidean geometry; Face recognition; Image detection; Image processing; Image resolution; Information processing; Internet of Things; Latency; Mobile applications; Mobile computing; Mobile devices; Motion perception; Object recognition; Optimization; Pattern recognition; Posture; Servers; Training; Wireless telephone software
• DOI: 10.1155/2022/7423411
• ISSN: 1687-5265; 1687-5273

The rapid development of Internet of things mobile application technology and artificial intelligence technology has given birth to a lot of services that can meet the needs of modern life, such as augmented reality technology, face recognition services, and language recognition and translation, which are often applied to various fields, and some other aspects of information communication and processing services. It has been used on various mobile phone, computer, or tablet user clients. Terminal equipment is subject to the ultralow latency and low energy consumption requirements of the above-mentioned applications. Therefore, the gap between resource-demanding application services and resource-limited mobile devices will bring great problems to the current and future development of IoT mobile applications. Based on the local image features of depth images, this paper designs an image detection method for athletes’ motion posture. First, according to the characteristics of the local image, the depth image of the athlete obtained through Kinect is converted into bone point data. Next, a 3-stage exploration algorithm is used to perform block matching calculations on the athlete’s bone point image to predict the athlete’s movement posture. At the same time, using the characteristics of the Euclidean distance of the bone point image, the movement behavior is recognized. According to the experimental results, for some external environmental factors, such as sun illumination and other factors, the image detection method designed in this paper can effectively avoid their interference and influence and show the movement posture of athletes, showing excellent accuracy and robustness in predicting the movement posture of athletes and action recognition. This method can simplify a series of calibration tasks in the initial stage of 3D video surveillance and infer the posture of the observation target and recognize it in real time. The one that has good application values has specific reference values for the same job.