基于双NiosⅡ的红外图像实时Otsu局部递归分割算法设计

Design of Infrared Image Real-time Local Recursive Otsu Segmentation Based on Dual NiosⅡ

针对传统Otsu局部递归分割方法很难实时实现的局限性,提出了一种适合FPGA(现场可编程门阵列)中N iosⅡ软核处理器实现的快速Otsu局部递归分割算法,该算法的思路是把一次Otsu分割得到的目标区域作为新的图像再进行一次Otsu分割,得到的结果作为最终分割阈值。利用并行N iosⅡ和VHDL实现的硬件加速逻辑协同设计保证算法的实时实现。实验结果表明,在不同的背景下,利用本文设计能够实时稳定地对目标分割提取,具有较好的鲁棒性。

In view of the limitations of traditional local zation is difficult, a fast local recursive Otsu＇ s segmentation in FPGA is proposed. The main technique of this algorithm is regarded as the whole image to be segmented at the next recursive otsu＇ s segmentation that real-time realialgorithm adapted to Nios Ⅱ software core realized consists in that the object area resulting from otsu otsu segmentation. Then the object areas resulting from otsu are the image of segmentation results. The design uses parallel Nios Ⅱ and hardware to realize realtime algorithm. The experimental result shows that, under different background, the algorithm in question is robust against background clutter and capable of realizing object segmentation and extraction.