摘要
提出了一种基于图像特征与面向对象BP神经网络的边缘检测方法。首先,在图像特征的选取上,充分考虑边缘和噪声的本质区别,构造具有较强抗噪能力的特征向量;然后用经人工处理的样本对BP网络进行训练,将训练后的网络用于图像的边缘检测。
A kind of edge detection method based on image features and object-oriented BP neural network is proposed in this paper. Firstly, the essential differences of noise and edge are fully considered in selecting the image features to form feature vector with better anti-noise performance, then a BP neural networks is trained with manual treated sample, finally the trained network is used to detect images edge.
出处
《计算机工程》
CAS
CSCD
北大核心
2002年第4期242-244,共3页
Computer Engineering
