基于BP神经网络和纹理特征的马尾松图像分割方法

Image segmentation of pinus massoniana based on BP neural network and texture feature

马尾松图像中树干部分的纹理可以很好地描述图像中的分割目标,表明神经网络可应用于该分类问题,如果给予较理想的训练样本,就可以得到较好的网络;再用此训练好的网络对输入图像进行分割就会得到较好的结果.结果表明这种方法效果良好.

The image texture of pinus massoniana trunks could be a good description of the target image segmentation. The result showed the neural network could be applied to this kind problem. If ideal training samples were provided, the result of segmentation could be better. The experiment results showed the effect of this method was better.