资源列表
foeu-multiobject
- 多目标检测，用到了马尔科夫随机场和高斯混合建模-multiobject detection
foeu-source-code
- 单目标检测，用到了马尔科夫随机场和高斯混合模型建模-single object detection
number
- 基于神经网络训练测试，能识别0-9个数字，字体可不同-Neural network training based tests can identify numbers 0-9, fonts can be different
image-scaling
- 介绍了用深度学习方法做超分辨率缩放问题，都要是2014年ECCV的两篇论文-deep learning super-resolution eccv2014 paper
mycnn
- 卷积神经网络识别字符的Matlab程序，包含所需的所有素材和自己改进的一部分代码-Convolutional neural network for handwriten digits recognition: training and simulation. This program implements the convolutional neural network for MNIST handwriten digits recognition, created by Yan
lanedetection
- 车道线检测c++源码，对图像进行预处理，包括二值化，边缘检测等，最后利用霍夫变换。- Lane detection c++ source code, for image preprocessing, including two values, such as edge detection, finally using Hof transform.
Sparse-Representation
- 稀疏表示用于人脸识别，采用SDR方法，文字发表在IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE顶级期刊，影响因子4.75.-Sparse representation for face recognition, the use of the SDR method, text published in the IEEE Transactions on pattern analysis and machine intell
texture
- 图像纹理分析综述PPT，介绍了灰度共生矩阵的概念及常用的提取方法-image texture PPT
Texture
- 基于灰度共生矩阵的图像纹理特征提取，所采用的默认为灰度图像，提取了0 45 90 135四个矩阵-texture feature based GLCM
Edge_Detection
- 多种边缘检测算法 可以进行对比选择，方便提取更好的边缘特征-A variety of edge detection algorithm can be compared to choose, easy to extract better edge features
DCT
- 这段程序是，J（p，q）代表的二值水印图像，二值水印图像的像素值只有0和1，所以if J（p，q）==0，a=-1 ，是如果水印图像像素值是0的话，让a=-1，当水印像素值为1的话，让a=1，我个人认为是为了提取时候好提取-DCT digital watermark embedding and extraction of
marching_cubes
- marching_cubes算法的C++实现-marching_cubes algorithm implemented in C++