期刊文献+

去模糊网络复层运动图像恢复算法仿真

Simulation of Deblurring Network for Multilayer Motion Image Restoration Algorithm
下载PDF
导出
摘要 复层运动图像由于模糊而导致细节纹理差,无法有效传递信息。为解决上述问题,基于AED健美操数据集,在数字图像处理的基础上,构建出一种去模糊网络与运动目标检测相结合的图像恢复算法,即DGV2-YOL3算法。算法首先对AED数据样本进行截取并预处理,利用下采样方法提升图像特征提取能力;然后基于四组通道采集图像特征,减少25%系统参数量,并利用细粒度模块对特征进行残差学习;接着对学习特征进行上采用融合处理,通过全局跳跃方法重建图像;最后基于DBSACAN算法,自适应规划图像锚点与IOU阈值,识别检测出运动类型,提升检测准确率。仿真主观分析表明,较其它算法相比,DGV2-YOL3算法处理后的图像,其纹理清晰,边缘显著,且“袜子”细节被恢复明显;客观分析结果显示,DGV2-YOL3算法具有最高的峰值信噪比与分类指标,较其它8类叠加算法相比,分别平均提升了13.48与5.01%,且数据处理效率排名第三,具有较高的实时性。综上所述,DGV2-YOL3算法有效的提升了模糊图像的边缘细节纹理恢复能力,具有较高的仿真研究价值。 Multi-layer motion images have poor detail texture due to blurring,which can not effectively transmit information.In order to solve the above problems,based on the AED aerobics data set and digital image processing,this paper constructs an image restoration algorithm which combines deblurring network with moving object detection,namely DGV2-YOL3 algorithm.Firstly,the AED data samples were intercepted and preprocessed,and the image feature extraction ability was improved by using the down-sampling method;Then the image features were collected based on four groups of channels,the number of system parameters is reduced by 25%,and the residual learning of the features is carried out by using a fine-grained module;and then the learning features are fused,and an image was reconstructed by using a global jump method;Finally,based on the DBSACAN algorithm,the image anchor point and IOU threshold were adaptively planned to identify and detect the motion type and improve the detection accuracy.The subjective analysis of simulation experiments shows that the image processed by DGV2-YOL3 algorithm has clear texture,prominent edges and obvious recovery of″socks″details compared with other algorithms;The objective analysis results show that DGV2-YOL3 algorithm has the highest peak signal to noise ratio(PSNR)and classification index,which are improved by 13.48%and 5.01%respectively compared with the other eight stacking algorithms,and the data processing efficiency ranks third,which has high real-time performance.To sum up,the DGV2-YOL3 algorithm effectively improves the edge detail texture restoration ability of the blurred image,and has high simulation research value.
作者 王峥 赵新辉 WANG Zheng;ZHAO Xin-hui(Physical Education College of Zhengzhou University,Zhengzhou Henan 450044,China)
出处 《计算机仿真》 2024年第5期264-269,共6页 Computer Simulation
基金 国家重点研发计划重点专项项目(2020YFC2006800) 河南省科技攻关项目(232102320309) 河南省哲学社会科学规划项目(2022BTY022) 河南省高等学校重点科研项目(22A890011)。
关键词 去模糊网络 图像恢复 运动检测 Deblurring network Image restoration Motion detec
  • 引文网络
  • 相关文献

参考文献10

二级参考文献46

共引文献38

;
使用帮助 返回顶部