基于LVQ神经网络脑部CT边缘检测方法的研究

An Edge Detection Method for Brain CT Scan Images Based on LVQ Neural Network

研究运用LVQ神经网络进行脑部CT图像边缘检测的方法,以提高边缘检测的准确度。使用传统的Sobel算法对脑部CT图像进行边缘检测,作为网络学习的教师信号,并将脑部CT图像的中值特征量、方向性信息特征量、Krisch算子方向特征量3项特征量作为LVQ神经网络的输入信号,进行网络训练后,再将训练好的网络进行边缘检测。在Matlab 2010平台下进行仿真对比,结果显示改进算法边缘检测结果与实际相符,比传统Sobel算法更具优越性。

An edge detection method for brain CT scan images using LVQ neural network is studied to enhance the accuracy of edge detection. First,the traditional Sobel algorithm is used to detect the edge of brain CT scan images as the teacher’s signal for network learning. Second,the value of the median,the direction of information characteristics,the direction of the characteristics of the Krisch operator of brain CT scan images are adopted as the input signal of the LVQ neural network to conduct network training. Finally,the trained neural network is used for edge detection. Comparison of computer simulations on M atlab 2010 shows that the improved method can achieve a credible result and has a better detection effect than the traditional Sobel algorithm.