基于CNN的滤镜信息检测

网络的高速发展和技术手段的不断进步使人们对照片的美化要求也在不断提高,使得拍照时各种各样的滤镜层出不穷。滤镜不仅能使图像更加美观,有时一些特殊的照片上加滤镜,会对人们产生误导并造成负面影响。基于现存的全图滤镜信息检测的算法较少,我们提出利用 CNN 卷积神经网络进行全图滤镜信息检测。最终经过实验验证检测取得了较好的效果,具有创新性、先进性等特点。

Study of the imaging property of a fluorescent confocal microscopy with a phase-only filter in an extended source

The phase information of an enlarged source is reconstructed with an annular two-zone phase-only filter in a fluorescent confocal scanning optical microscope for resolution improvement. The dependences of its resolution on the source size and on the phase transmission of the outer annular zone of the falter are investigated theoretically by use of its three-dimensional optical transfer function （3D OTF ）. The increased source size and the required phase value of the outer annular zone of the phase-only falter for an optimal 3D OTF of the optical system are presented.