多尺度图像边缘检测技术在车辆识别中的应用

指出了图像边缘检测的定义及现有算法的不足,并对Canny边缘检测离散准则及该准则下的最优线性滤波器进行了论述。根据边缘检测离散准则,利用数值方法求出该准则下边缘检测的最优线性滤波器及对应的平滑算子,将其与Canny边缘检测技术相结合进行边缘检测,检测出不同尺度下多幅边缘图像,并根据小尺度下图像边缘细节信息丰富、边缘定位精度高,大尺度下图像边缘稳定、抗噪性好等特点,将检测到的多尺度边缘进行融合,得到抗噪性好的精确的单像素宽边缘;同时将该方法应用于汽车特征提取,检测到完整连续的汽车边缘图像,为车型识别提供条件。试验结果表明:该方法不仅能准确检测出图像边缘,而且能有效地抑制噪声,是边缘检测的实用方法。

The Technique of Multi-scale Image Edge Detection Applying to Vehicle Recognition

The definition of image edge detection is proposed,disadvantages of traditional methods are indicated.Discrete criteria of Canny and the optimal filter of the criteria are discussed.According to discrete criteria of edge detection,numerical method is used to get the optimal linear filter and smooth operator.Then the smooth filter operator is designed and combined with Canny algorithm to obtain multiple edge images under different scales,and with full use of the multi-scale character of the edge information which has rich details and more accuracy with small scales and higher signal-to-noise-rate with large scales,the detected multi-scale edges are synthesized to obtain accurate single pixel wide edges.Further more,this method is applied to vehicle features detection,and continuous vehicle edges are obtained which can provide effective information for vehicle type recognition.The experimental results demonstrate that the method can not only detect image edges precisely,but also restrain noise effectively,and is a practical edge detection algorithm.