摘要
针对图像语义分割不能在提高分割速率的同时改善分割结果的问题,基于深度学习和图像去噪的方法提出一种对图像进行预处理再输入到图像语义模型中解决方法。该方法通过使用大量数据集训练网络模型,再将预处理图像进行去噪之后输入到训练好的网络模型中,利用SegNet网络作为分割网络模型完成道路图像的实时语义分割和基于差异系数的稀疏自适应度去噪算法完成对图像的预处理。实验证明,对图像进行去噪处理之后再分割能够再不增加计算量的基础上有效的提高分割精度。
For the problem that image semantic segmentation can not improve the segmentation result while improving the segmentation rate,a method based on deep learning and image denoising is proposed to preprocess the image and input it into the image semantic model.The method trains the network model by using a large number of data sets,then denoises the preprocessed image and inputs it into the trained network model,and uses the SegNet network as the segmentation network model to complete the real-time semantic segmentation of the road image and the sparseness based on the difference coefficient.The fitness denoising algorithm completes the preprocessing of the image.Experiments show that the image can be denoised and then segmented to increase the segmentation accuracy without increasing the amount of computation.
作者
张蓉
赵昆淇
顾凯
ZHANG Rong;ZHAO Kunqi;GU Kai(School of Electronic Information,Jiangsu University of Science and Technology,Zhenjiang 212000)
出处
《计算机与数字工程》
2020年第7期1772-1775,1803,共5页
Computer & Digital Engineering
关键词
图像语义分割
卷积神经网络
SegNet
去噪
image semantic segmentation
convolutional neural network
SegNet
denoising
