快速Hough变换(FHT)在实时识别目标系统中的应用

Application of the Fast Hough Transform in Real-Time Object Recognition System

Hough变换(HT)是一种在噪声或退化图像中检测直线的强有力工具。FHT采用了分层技术和K-树结构,从而大大提高了检测速度,减少了内存量。本文把FHT用于实时识别目标图像系统中,完成了模拟实验,实现了在复杂背景中实时识别目标,同时输出目标位置。由于利用了HT,使得整个系统对于有噪声、目标残缺、背景复杂的目标图像具有较强的识别能力。在目标样板选定之后,系统对目标的识别将不受背景变化的影响,并且具有目标旋转不变性。

Hough transform is powerful for detecting straight, line in noisy image or degeneration image, Fast Hough transform(FHT) adopts a hierarchical technique and a k-tree data structure. It leads to a reduction in both computation and storage. In this paper, a technique applied FHT to real-time object recognition system. We now accomplished the computer simulation for the real-time object recognition system is described. The system can output the position of the object in an complicated background. Due to the HT, the system is very effective in recognizing object in an image with noise, gaps and complicated background. After determining a mask, the object recognition in this system is not affected by the background variation, and it has the feature of object rotation invariant.