Microservice Driven Parallel Framework for Image Collection and Processing in the Large-scale Laser Beam Collimation System

Abstract

Laser beam collimation involving image collection and processing is an essential task in the high-power laser experiment. With the development of worldwide high-power laser facilities, both the type and the amount of image collection devices have increased, which contributes to a large-scale heterogeneous image collection system and a highly concurrent image data environment. It is challenging our current image collection and processing system, and requiring a more efficient control pattern. Parallelism by utilizing the multi-thread technology is an effective solution, while a big obstacle is that image collection devices can not be parallel driven, thus we obtain the pseudo parallelism. To address this issue, giving the existing microservice based distributed architecture utilized by the laser beam collimation system, we equip each of these devices with an embedded computer. We conduct the co-design of hardware and software for each image collection device, such that each embedded computer serves as an independent and parallel servant of image collection and processing. This parallel framework is expected to reduce the time consumption of image acquisition and processing. What's more, hardware customization can be further implemented.