source

卷积神经网络（Convolutional Neural Network）是一个专门针对图像识别问题设计的神经网络。它模仿人类识别图像的多层过程：瞳孔摄入像素；大脑皮层某些细胞初步处理，发现形状边缘、方向；抽象判定形状（如圆形、方形）；进一步抽象判定（如判断物体是气球）。(Convolutional neural network is a neural network specially designed for image recognition. It imitates the multi-layer process of human recognition image: pupil ingestion of pixels; some cells in the cerebral cortex preliminary processing to find shape edges and directions; Abstract judgment of shape (such as circles and squares); and further Abstract judgment (such as judging objects are balloons).)

详细介绍了常用的四中粒子重采样算法,并给出了他们的性能比较,对研究改进型粒子算法很有帮助,减轻很多工作负担(Details of the four commonly used particle resampling algorithm, and gives the performance of their comparison, the study of modified particle algorithm useful to reduce the burden of a lot of work)

包含了21篇二维最大类间方差（Otsu）阈值分割方法的论文，(Contains 21 article 2 d between the most kinds of variance (Otsu) threshold segmentation method papers)

很好的模糊聚类算法 用MATLAB编写的(Good fuzzy clustering algorithm using MATLAB prepared)

快速非局部均值，用来消除噪声，效果良好，适合于细节较多的图像(Fast nonlocal means is used to eliminate noise and achieve good results, which is suitable for more detailed images.)

这是一个简单图像处理的GUI程序。实现的功能有：直方图统计，直方图拟合，边缘提取和自定义模板算子处理。界面采用的是ENVI 4.7的风格。(This is a simple image processing GUI program. Implemented features are: histogram, histogram fitting, edge extraction and custom template operator handling. Interface using the ENVI 4.7 style.)

基于小波变换的利用阈值法迭代法的图像分割技术(Wavelet Segmentation of the iteration method)

说明：  用matlab语言开发的图像平移程序，包含测试图片(Matlab language used to develop the image translation process, including test picture)

实验图像数据曲线描绘刻画，进行数据拟合：高斯拟合，平滑处理(gaussian nihe)

运用harr特征+SRC（稀疏表示）分类实现的一种车辆检测方法，文件中提供了训练和测试车辆图片。由于时间原因，所用haar特征没有优化，维度过高，导致滑窗框图过慢，本代码只输出效果统计数据，以供大家参考学习稀疏表示在车辆检测中的应用。(Using harr feature+SRC (sparse representation) classification to achieve a vehicle detection method, the paper provides a training and test vehicle picture. Due to time reasons, the use of haar feature is not optimized, high dimension, resulting in sliding sash figure is too slow, the effect of the code only output statistics for your reference learning sparse representation in the vehicle detection.)