Respiration detection is often used to analyze the physiological state of the human body, sleep quality and other applications. For traditional respiration detection, contact equipment must be used, which is easy to bring discomfort, and is easy to fall off due to body movement, which is inconvenient. These problems can be avoided through non-contact detection by image sensors. This article will show you Arrow Electronics' breath detection solution as reference for your relevant product development.
Non-contact detection has better performance characteristics
Breathing is an important physiological signal of human beings. By examining the breathing process, it can reflect the health state of the subject. In conjunction with analysis of the algorithm, the physical condition and sleep quality of the subject can be analyzed. Breathing can also be applied to detect the mental state of a driver when driving, diagnose apnea, or detect the sleep state of the elderly and infants, so as to find any sleep apnea condition as early as possible and avoid accidents.
Traditional breathing detection device must be connected to the subject, which will cause discomfort, and it is easy to fall off due to turning over and moving too much, resulting in detection failure. The new respiratory detection device can capture the image of the subject through the image sensor, and then compare and analyze it through an artificial intelligence (AI) algorithm, which can detect the respiratory state of the subject and avoid the shortcomings of contact equipment.
Respiratory frequency detection by AI algorithm
Using the latest AI video amplification technology, breath detection solution introduced by Arrow Electronics can magnify the micro-movement of the human body. For example, the chest and abdomen will fluctuate up and down due to breathing, which can be used to calculate the breathing rate. In cooperation with a third-party research institution in Hong Kong, this solution uses OpenCV and TensorFlow Lite library to carry out AI algorithm, and works with different supplier products, including NXP’s AI platform, ON Semiconductor’s camera, Murata’s Wi-Fi module, etc., to form a complete breathing detection solution.
This breath detection solution uses a camera to video the subject, and carries out image processing of motion enlargement to judge the characteristics of breathing rhythm. It can enlarge the up-and-down micro-movement of the chest and define the scene state at the same time. When "scene unstable " occurs, the detection will be stopped until the image is stable and there is no background interference, or until the human body is stable, and the threshold parameters can be adjusted to correspond to different application environments.
When the subject's image is stabilized, the target human body will be searched by an AI algorithm and only the motion state of the region will be calculated. The signal-to-noise ratio of respiratory rate detection is improved and the detection accuracy is enhanced. The enlarged image of human body motion is converted into a graph to view the motion level in real time, and the respiratory frequency is calculated according to the threshold level. In addition, parameters can be modified to fine-tune the performance of other scenarios.
The hardware of this breath detection solution uses NXP i.MX 8M mini computing platform, which is composed of quad-core ARM Cortex-A53, a Cortex-M4 and GPU. It supports AI TensorFlow Lite and OpenCV engine, and inserts RGB camera daughter board plugs on I.MX development kit. The image signal is processed by an AI image algorithm to measure breathing frequency. The result can be displayed on the display screen supporting HDMI interface, or sent to the cloud for recording through WiFi wireless transmission for future viewing.
Conclusion
The shortcomings of traditional contact breathing detection devices can be improved by non-contact image detection technology. Combined with an AI algorithm, the detection accuracy can be greatly improved. It is one of the best breathing detection technologies at present. Arrow Electronics provides a complete software and hardware reference design for respiratory detection applications for reference by relevant product developers to speed up product development and bring products to market as soon as possible.