首页> 外国专利> Computer realization method using convolution neural network, apparatus for generating composite image and computer program product

Computer realization method using convolution neural network, apparatus for generating composite image and computer program product

机译:计算机实现方法使用卷积神经网络,用于生成合成图像和计算机程序产品的仪器

摘要

A computer realization method using convolution neural network is provided.The computer implementation method includes processing the input image by a convolution neural network and generating an output image that includes the content features of the input image in which the characteristic features of the fine image are fused.The convolution neural network includes a feature extraction subnetwork, a fusion portion, and a decoding sub network.The process of processing input images by convolution neural networks isFeature extraction sub networkThe characteristic features of the image quality are extracted.Generating multiple quality feature maps andFeature extraction sub networkExtract the content feature of the input imageCreating multiple content feature mapsMultiple feature feature maps and multiple feature feature maps are fused using fusion.Generating multiple output feature mapsBy decoding subnetworkReconfiguring multiple output feature mapsGenerating output images and.[Orig] Figure 6A
机译:提供了一种计算机实现方法,使用卷积神经网络。计算机实现方法包括通过卷积神经网络处理输入图像并生成包括输入图像的内容特征的输出图像,其中细图像的特征融合。卷积神经网络包括特征提取子网,融合部分和解码子网络。通过卷积神经网络处理输入图像的过程ISFeature提取子网络提取图像质量的特征特征。生成多个质量的特征映射和Feature提取子Networkextract的内容特征模拟多内容Feature MapSmultiple特征映射和多个特征要素映射使用Fusion融合。生成多个输出特征MapsBy解码子网,SubnetworkReconfiguring多个输出特征MapSgenerating输出图像和。[ogg] Fimor e 6a.

著录项

相似文献

  • 专利
  • 外文文献
  • 中文文献
获取专利

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号