Comparison of CPU and GPU-based Image Processing Performance in OIS Inspection Equipment

This paper presents a novel stereo visual simultaneous localization and mapping Before installing optical image stabilization (OIS) systems in devices such as smartphones, an inspection process is conducted on the OIS systems. During this process, images of the OIS system are captured using a camera, calculations are performed, and the results are used to verify regular operation. In this calculation process, cameras with higher resolution and frames per second provide more precise and accurate results as the processing speed increases. In this study, we enhance performance by reducing image processing time through parallel computation using a GPU while constructing an environment that can utilize both integrated and individual GPUs to secure versatility in equipment usage. Additionally, we ensure scalability for potential future technological applications as needed.