A Camera Placement System for Motion Analysis and Object Recognition: System Assessment by Simulations and an Experiment

In manufacturing plants, highly technical processes are frequently performed. Hand operated tasks for soldering electronic components to circuit boards are important processes that directly affect the performance and reliability of the products. In many operations performed by people using tools, worker safety is required to prevent burns, cuts and other injuries. Therefore, the workers movement data should be collected and analyzed in order to provide safe environment. Non-contact sensing systems include methods that use cameras to capture images of workers or products and use skeletal estimation and object recognition to detect dangerous movements of workers and product conditions. For carrying out the data analysis, the worker and object should be within the camera view-angle. Therefore, the camera placement is crucial for capturing workers movements and object state accurately. In this paper, we propose a camera placement system in order to carry out motion analysis and object recognition. From the simulation and experimental results, we found that by deciding the camera placement and angle of view, the proposed system can be used for motion analysis and object recognition. Also, the proposed system is an effective system for improving the recognition accuracy and detection rate of humans and objects.