Research On Collimated Image Embedded Processing Platform Based On Heterogeneous Multi-core Processor

Benefit from the development of CCD sensor technology, the SG-? laser facility is capable of deploying a large number of high-resolution digital CCD cameras in its automatic alignment system, thereby performing high-precision laser alignment. However, a large number of high-resolution CCD cameras means that a large amount of image data. Especially when SG-? uses a centralized image processing system, a large amount of image data increases the payload of network transmission, storage and image processing.But in fact SG-?'s automatic laser alignment system only requires the characteristic data of an image, such as contour, centroid, geometric center for closed-loop control. The data is much smaller than the original image data. If these features can be extracted in the CCD locally, that can avoid a lot of raw image data, thereby eliminating payload of storage and image processing.Therefore, based on the problems of SG-? centralized image processing mentioned above and the characteristics of alignment system's image processing, this thesis conducted a research on embedded image processing platform for laser alignment based on heterogeneous multi-core processors.First, this thesis analyzed the characteristics of the current SG-? centralized image processing system, maximizing the compatible of the current system framework designed a distributed embedded image processing systems based on embedded devices. By using the CCD camera embedded in the local device performs image processing, image characteristic data extracting that alignment system needed to avoid a large number of redundant data transmission.Secondly, this paper analyzes the current embedded image processing, integrated performance and power consumption considerations, using heterogeneous multi-core processors for image processing. After evaluated several heterogeneous multi-core processors, choosing DM3730 and DaVinci software framework designed multi-core software framework and boot solution. And conducted research studies multicore communication technologies. The embedded image processing platform designed base on these studies.Then, based on processor DM3730 development board this thesis designed and implemented embedded image processing platform prototype system for verifying the DM3730 heterogeneous multi-core processor performance of alignment image processing.Finally, the test of prototype system verified the embedded hardware meets time requirements. So finally verified the heterogeneous multicore processors image embedded processing platform designed in this thesis can be used to solve the problem of facing a large number of image data transmission,storage and image processing, when the SG-III or future larger laser device using centralized image processing solution.