基于TMS320C6678的细胞图像识别系统并行实现方法

针对医学显微图像自动分析系统在应用中的实时性需求,根据显微镜细胞图像识别原理和多核流水设计准则,分析了细胞图像识别算法实时性,提出了一种基于TMS320C6678显微镜细胞图像识别并行流水实现方法。该方法实现了任务级并行流水,提高了系统的鲁棒性,达到了系统稳定和性能加速的目的。实验结果表明,该方法是可行、有效、可靠的,并且基于TMS320C6678的显微镜细胞识别系统的实时处理能力有很大提升。

Parallel realization of cell image recognition system based on TMS320C6678

Aimming at the real-time requirement of the medical microscopic image recognition in equipment application, the real-time performance of cell image recognition algorithm is analyzed according to the cell image recognition theory and multi-core DSP design guideline,and a parallel implementation method of microscope cell image recognition algorithm based on TMS320C6678 is put forward. The method implemented task-level parallel pipeline,improved the robustness of the system, and reached the stability and performance acceleration of the system. Experiment results show that the method is feasible,effec-tive and reliable. The real-time processing capability of microscope cell image recognition system based on TMS320C6678 was improved much more.