noise genre

提取一幅图像的LBP特征，输入为一幅图片，输出为一个256维的行向量，代表图像的LBP特征。(The LBP feature of an image is extracted, and input is a picture. The output is a 256 dimensional row vector, representing the LBP characteristic of the image.)

说明：  灰度图像的边界跟踪的方法，先用微分算子计算出图像的梯度图。然后再应用此算法。(Gray-scale image of the border tracking methods, first calculate the differential operator of the gradient map images. And then apply this algorithm.)

数字图像处理课程期终项目，对图像进行放大，并进行滤波，达到一定的信噪比。(The final project of the digital image processing lesson, which is programmed for enlarging a image into demanding size.)

下面是利用卡尔曼滤波的方法，实现雷达对目标的跟踪:一目标沿水平方向运动，起始位置为（－2000米，1000米），运动速度为15米/秒，扫描周期T＝2秒， 米，采用蒙特卡洛方法对跟踪滤波器进行仿真，仿真次数为100次。(Below is the use of Kalman filtering method, to achieve the target tracking radar. Parameter :: a target in a horizontal direction, the starting position (2000m, 1000m), the motion speed of 15 m/s, the scan period T = 2 seconds, rice, Monte Carlo method using the tracking filter simulation, simulation 100 times.)

说明：  将ply格式的3D物体文件导入Matlab中，并且可以把Matlab中的Patch object导出成ply格式。(Ply format to 3D objects in the file into Matlab, and Matlab can be derived in the Patch object into a format ply.)

non-local means denoising 自己实现的非局部均值图像去噪的原始算法，对理解算法原理有一定帮助(Non-local means denoising： an implementation of the non-local means denoising of the original algorithm, some help understanding algorithm)

运用openCV编写的C++图像色彩转移程序，原理是先转化到Lab图像，再保持L不变调整图像的a、b分量的均值跟方差。(Using openCV prepared C++ image color transfer process, the principle is first converted to Lab image, and then adjust the image to keep L constant a, b, with the mean variance component.)

说明：  阴影3D重建,对于人脸有效，基于yale2BDB(SFS, effective for face, based on yale 2 B database)

利用灰度积分投影直接对人脸图像进行检测和眼睛定位是一种常用的算法,但是直接采用该算法会受到背 景、特征等因素的影响,识别准确率较低。提出了一种基于最大类间方差阈值和区域膨胀相结合的检测与定位算法。该算 法首先计算最大类间方差设置阈值,把灰度图像转换为二值图像并检测出人脸区域,然后通过对该人脸区域中的连通区域 进行膨胀及连通性处理,精确定位眼睛坐标。实验表明,此算法可靠,具有较好的识别效果。(face recognized)

说明：  关于数字图像一些典型隐写的隐写分析。隐写、分析和提取的讲解(Steganalysis of Some Typical Steganography in Digital Images)