基于双边缘检测的目标定位技术

为了提高目标定位的准确性，在传统的高斯差分边缘检测和Sobel算子边缘检测的基础上，提出将二者相结合的双边缘检测算法。对原始图像进行2次标准差不同的高斯模糊求高斯差分来代替原本边缘检测图像预处理中的高斯平滑过程；对实验结果较好的高斯差分参数，进行Sobel边缘检测并进行目标定位。实验结果表明，在算法时间开销增加较少的情况下，使用高斯差分算子与Sobel算子相结合进行边缘检测的结果优于Sobel算子单独检测，检测效果较好。

Target Location Based on Double-edge Detection

In order to improve the accuracy of target location, based on the traditional difference of Gaussian edge detection and Sobel operator edge detection, this paper puts forward a double edge detection algorithm which combines the two. We use the difference of Gaussian to replace the Gaussian smoothing in image preprocessing, choose better results of difference of Gaussian parameters, use Sobel edge detection after the difference of Gaussian process, and find the target location. The experimental results show that combining difference of Gaussian with Sobel operator do not increase much running time overhead, and have better detection result than the Sobel test alone.