NSST_PAPCNN_Fusion

利用全卷积神经网络，实现图像的语义分割，基于tensorflow的keras可以直接运行(Using the full convolutional neural network to achieve semantic segmentation of images, keras based on tensorflow can be run directly)

3种不同的分水岭分割方法以及三种方法的比较(3 different methods of watershed segmentation and Comparison of Three Methods)

说明：  结合Retinex和CLAHE图像去雾，Retinex有ssr，msr，msrcr。(Combine Retinex and clean image to remove fog)

基于全变分（TV）的多光谱遥感图像恢复。当多光谱图像中一个通道模糊时，利用其它通道的图像特征辅助进行去模糊或反卷积，效果不错！！附带有对应的文章做参考。(This is a new image restoration or deconvolution method, which uses one of clear band image to constraint Total Variation de-blurring of degraded band image in multispectral image. It tested with different noise with different regularization parameters. It also compared with otherhods.)

在matlab环境中用快速傅里叶变换实现多目标追踪，计算的非常快(Matlab environment using the fast Fourier transform to achieve multi-target tracking, very fast calculation)

python语言，使用卷积神经网络的特征做图像检索(image retrieval using CNN)

本文中我们提出一个新颖的算法分析和 估计深度从单个图像捕捉到一个编码的孔径 相机。这是一个具有挑战性的问题,需要 新的工具和调查,而多视点重建。 (In this paper we present analysis and a novel algorithm to estimate depth from a single image captured by a coded aperture camera. This is a challenging problem which requires new tools and investigations, compared with multi-view reconstruction.)

对理想的心音信号加入噪声，仿真成实际带噪声的心音信号，然后对其进行维纳滤波程序(The ideal heart sound signal added noise, simulation into actual heart sounds with noise signal, and then its Wiener filtering program)

全部都是图像处理的最新文献，与激光雷达有关，利用激光与摄像头结合，实现主动驾驶(image processing)

通过GAN和WGAN在python的tensorflow平台上实现图片生成，与真实图片基本一致。(Through GAN and WGAN, the image generation is realized on the tensorflow platform of python, which is basically the same as the real picture.)