As the impact of population aging becomes increasingly serious, more problems emerge with regard to medical and long-term care for the older individuals, which require solution. In long-term care, accidental falls frequently occur and are a significant cause of serious injuries and bed confinement due to leg fractures. Older people are especially prone to fall out of bed when they try to get up on their own. To reduce the occurrence of such accidents, one important solution is to detect the getting-up behavior of older individuals and to alert nursing staff to come and check. Although many studies have proposed various solutions such as the use of wearable devices and vision-based sensors, there are many issues in practical application. The complexity of device installation, high initial cost, and maintenance problems have restricted most care facilities to using fall prevention systems with mediocre results, such as pressure pad sensors. In this paper, we propose a fall prevention system based on a single camera (with infrared function) and infrared reflective sheets (IR sheets). The system detects the IR sheets placed on the shoulders of an individual and obtains position data. The relative positions of the IR sheets can be used to identify the state of the person on the bed. To improve the identification ability of the system, we propose to define an identification area. To identify the state more accurately, we propose to establish sub-areas within the identification area. We conducted experiments by recruited 19 subjects. The data of 6 subjects were used to construct the sub-areas. The other 13 subjects participated in testing the ability of the system in identifying the various states of the person in bed. Compared with the performance of other studies, our experimental results demonstrate that our system has a high identification rate, in addition to being low-cost and easy to set up.