SSDA并行算法在无纸化阅卷中空白题识别的应用

SSDA parallel algorithms in the paperless mark of blank question identify

图像匹配是图像处理领域的一项重要技术。针对无纸化阅卷系统的主观题评阅,提出了一种结合自适应阈值SSDA(序贯相似性检测算法)和多核多线程并行技术的图像匹配方法,由电脑自动完成主观题空白题的评阅,以提高评阅工作效率。通过螺旋分配多核处理器资源,在多核处理器的每个核心上并行运行SSDA图像匹配,按行列顺序处理待匹配区域,每个核心的匹配结果不断更新阈值,达到提高匹配速度的目的。最后进行实验仿真,可见,与传统的串行算法相比较,本算法能够满足空白题识别中图像匹配的实时性要求,并且有较好的识别效果。

Image matching is an important technology in image processing field. For subjective questions review of paperless scoring system,in order to improve review efficiency we propose an image matching method combining adaptive threshold SSDA（ Sequential Similarity Detection Algorithm） technology with multi-core multi-threaded parallel,thus,the computer can automatically complete review of blank questions of subjective questions. By assigning multi-core processor resources,spiral on each core of multi-core processors run in parallel SSDA image matching,according to the order processing for matching area,each core of matching results update threshold,to achieve the goal of improve the matching speed. Finally do the simulation experiment. Compared to traditional serial algorithm,this paper proves that it meets the real-time requirements of image matching in blank problem identification,and has better recognition results.