摘要
利用CP神经网络分别实现了对二值图像和灰度图像的边缘检测。利用神经网络提取二值黑白图像的边缘,对于灰度级为256的灰度图(灰度值在8个位面上分别为0或1),采用已学习好的二值神经网络在8个面上分别进行边缘检测,然后综合各个面的检测结果,解决了直接用灰度图学习造成的训练样本数量过大而难以收敛的问题。试验结果表明,该方法得到的边缘图像边界连续性较好,边界封闭性好,而且对于任何256级灰度图的检测都可以得到很好的效果。
A method of edge detection of binary images and grayscale images using CP neural network is proposed. CP neural network is implemented to detect the binary images. A grayscale image with it is pixel value arranged from 0 to 255 can be divided into 8 binary planes. The edge of grayscale images then can be detected through synthesizing the edge of each binary plane using the trained neural network. This method avoids the difficult convergence of overabundance samples if grayscale images are directly used to train the neural network. Experiment proves that the continuity of edge obtained with this method is better and the edge closed well, and that good result can be obtained when any 256level image is tested.
出处
《四川大学学报（工程科学版）》
EI
CAS
CSCD
2003年第3期93-96,共4页
Journal of Sichuan University (Engineering Science Edition)
基金
四川省科技厅重点项目基金资助项目(00C084)
