准确性和实时性是影响目标检测性能的两项重要指标。为了在多种复杂场景中进行快速精准的目标检测, 提出了一种融合超复杂傅里叶变换(HFT)显著性计算、高效子窗口搜索(ESS)和大津法(OTSU)的精准目标检测算法。该算法首先将输入图像进行HFT显著性计算, 得到显著图, 然后在显著图的基础上通过高效子窗口进行全局搜索并初步定位目标, 最后利用大津法分割子窗口范围内的图像, 并根据子窗口边界的像素值优化子窗口的位置, 实现精准的目标检测。实验结果表明, 新算法在保证运算速度的同时, 显著提高了检测精度。

Accuracy and real-time performance are two important indexes of target detection.In order to ensure that the target is quickly and accurately detected in a variety of complex scenarios,an accurate target tracking algorithm is proposed using Hypercomplex Fourier Transform (HFT) saliency calculation,Efficient Sub-window Search (ESS) and OTSU algorithm.First,the input image is preprocessed by the HFT algorithm to obtain the saliency map.Then the target is searched globally and preliminarily located by using the efficient sub-window based on the saliency map.Then the image within the scope of the sub-window is segmented by using OTSU algorithm.Finally,the position of the sub-window is optimized according to the pixel value on the edge of the sub-window,so as to ensure accurate target detection.The experimental results show that the algorithm greatly improves the real-time performance of detection without significant increase of computing time.