Table and Format Circular Saws are still one of the most dangerous machines at all. An analysis of almost 600 accidents in joiner’s and carpenter’s workshops shows that less experiences users are endangered particularly. We developed a computer based approach to detect body parts in video image sequences to avoid accidents in future.
Although there have been lots of efforts in the past to improve the safety at work, none of the new concepts has gained acceptance. The reasons for it are the bad discrimination between hands or fingers and wood. The analysis of the accidents shows that the blade protection cover must be integrated into a new concept to keep the hands away from the dangerous zone of the machines.
Coarse-to-fine Strategy: A video-based sensor has been developed which is able to register body parts within a defined dangerous zone in the shortest possible time. For the implementation of the video-based detection process, a course-to-fine strategy is followed. This means that beginning with the head and shoulder area of the person currently standing at the machine, a step-by-step detection and tracking of the arms, hands and –finally – fingers will be made using a specially-developed process of analyzing video signals. The application of the coarse-to-fine strategy ensures that at any time instance information about the dangerousness are available.
Image sequences of a worker are taken at a circular saw applying currently modest camera hardware and transmitted to a PC. Subsequent an adaptive foreground-/background segmentation process and further pre-processing steps, motion cues, edge and color information are derived and combined. Within the framework of the application of the coarse-to-fine strategy detailed edge information is analyzed which is provided by a steerable filtering scheme.
Transfer of the Applicability: The developed methodology should serve as a generic example for the detection of persons and body party generally in dangerous zones. Therefore, in further steps of the development process the developed approaches will be adapted to other machine classes, such as metal working machines, sewing machines, and industrial robots.