Login| Sign Up| Help| Contact|

Patent Searching and Data


Title:
動きボケ除去装置及びプログラム
Document Type and Number:
Japanese Patent JP7219727
Kind Code:
B2
Abstract:
To provide a motion blur removal device capable of efficiently removing a motion blur in an image while controlling a calculation amount.SOLUTION: Provided is a motion blur removal device comprising at least two network sections having a mutually common network structure, wherein utilizing a convolutional neural network obtains an output image in which a motion blur removal is performed to an input image by each of resolution scales different from one another. The network sections comprise an efficient space pyramid structure (E34, E25, E14) for extracting and integrating at least two features for space scales different from one another, as the network structure responsible for processing procedures until obtaining the output image from the input image.SELECTED DRAWING: Figure 2

Inventors:
Xu Jianfeng
Kazuyuki Tasaka
sign army
Application Number:
JP2020003972A
Publication Date:
February 08, 2023
Filing Date:
January 14, 2020
Export Citation:
Click for automatic bibliography generation   Help
Assignee:
KDDI CORPORATION
International Classes:
G06T7/00; G06T5/00
Domestic Patent References:
JP2017187954A
Foreign References:
KR1020190067111A
Other References:
Xin Tao et al.,Scale-Recurrent Network for Deep Image Deblurring,2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition,米国,IEEE,2018年06月23日,pp.8174-8182,https://ieeexplore.ieee.org/document/8578951
Sachin Mehta et al.,ESPNet: Efficient Spatial Pyramid of Dilated Convolutions for Semantic Segmentation,Computer Vision and Pattern Recognition,米国,2018年07月25日,pp.1-29,https://arxiv.org/abs/1803.06815
Attorney, Agent or Firm:
Kiyotaka Sakamoto
Toshiji Tanabe