Open AccessResearch on Image Segmentation based on Full Convolutional Neural Network
Author(s) -
Yanan Guo,
Zhihua Zhao,
Zuohong Wu
Publication year - 2020
Publication title -
iop conference series. earth and environmental science
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.179
H-Index - 26
eISSN - 1755-1307
pISSN - 1755-1315
DOI - 10.1088/1755-1315/440/5/052065
Subject(s) - artificial intelligence , image segmentation , segmentation based object categorization , computer science , scale space segmentation , segmentation , pattern recognition (psychology) , convolutional neural network , minimum spanning tree based segmentation , robustness (evolution) , computer vision , region growing , biochemistry , chemistry , gene
Image segmentation technology is an important branch in the field of computer vision. Image segmentation is the basis of image analysis processing, and image segmentation effect directly affects image subsequent processing. Aiming at the problem that the segmentation method based on traditional machine learning is not accurate, the edge information is lost and the robustness needs to be improved, this paper proposes an image segmentation algorithm based on improved full convolution neural network. The algorithm utilizes the better feature extraction ability of the deep learning model and the sensitivity of the cluster segmentation to the edge information, and further assists the segmentation with the Ncut algorithm. The experimental results show that compared with the traditional convolutional neural network image segmentation algorithm, the algorithm finally achieves higher segmentation accuracy. From the overall analysis, the segmentation method proposed in this paper is better than other methods.