GPExpand

研究生机器视觉课程的大作业 运动目标检测 (moving target detect)

很好用的三维重建软件，VisualSFM软件允许上传一系列图像，它从这些图像中找到每一个图像的特定特征，利用这些特征信息重建出3D模型的稀疏点云，而后还可进行稠密点云重建。(The VisualSFM software, a good 3D reconstruction software, allows uploading a series of images. It finds the specific features of each image from these images, and USES these feature information to reconstruct the sparse point cloud of the 3D model, and then it can reconstruct the dense point cloud.)

MNIST数据集中图像数据文件， 60000个训练集(The MNIST dataset image data files, 60000 training set)

图像分解算法，四种方法可选，包括threshold, K means threshold, region growing以及gradient based segmentation(Image decomposition algorithm, four methods available, including the threshold, K means threshold, region growing and gradient based segmentation)

详细介绍PCNN算法用于lena图像分割的实例，并用matlab实现(Detailed the PCNN algorithm for instance lena image segmentation and implemented using matlab)

图形处理的部分，能进行图像的中值过滤处理(To the median filter for image processing)

这里包含了多种图像倾斜角度矫正的方法 其中有huogh变换 radon变换 还有常规的线性检测角度 然后矫正它(This contains a variety of image angle correction method which huogh transform the linear radon transform routine testing point of view there and then correct it)

自适应直方图均衡算法的实现，即AHE算法，并提供样图进行对比。(Adaptive histogram equalization algorithm to achieve, namely AHE algorithm and provide comp image comparison.)

这篇文章是一种特征降维的方法，不过跟一般的计较离散度的方法不同，这个特征变换的准则是：尽可能地保留原始样本的熵(Kernel Entropy Component Analysis Jenssen, R Pattern Analysis and Machine Intelligence, IEEE Transactions on Volume: PP , Issue: 99 )

Ocr related documentataion