【摘要】 人耳识别作为新的生物特征识别技术,首先要解决作为基础的边缘检测和特征提取等图像处理方面的问题。传统边缘检测方法大多以各种微分算子为基础,结合模板及门限、平滑等手段提取边缘,或以传统微分算子为基础的改进算法,但都或多或少存在噪声敏感性较大或边缘保留不完整的缺点。论文采用二维非张量积样条小波的方法用于边缘检测,可以完整探测、准确捕捉到边缘点的位置,并且具有滤波作用,尺度j的自由选择可以有效提高边缘检测的效果。更多还原

【Abstract】 As a novel biometric authentication technology,human ear recognition need s to solve the problems of image processing such as edge detection and feature extraction.Most of the methods of classical edge detection are either based on various differential operators combined with the methods of threshold ranking,smoothing or template matching,or some optimized arithmetic based on the classic differential operators.The application of the above methods has some deficiencies such as sensitivity to yawp,loss of edge detection.In this paper,the application of two-dimensional non-tensor product spline wavelet can detect and catch accurately the position of edge points,and has satisfied filtering results.The optional choice of scale-j can raise the effectivity of detecting edge.更多还原