[图形图像处理] CamShiftDemo
camshift跟踪算法，可以自动确定跟踪目标的大小，中心，方向，运算速度很快，满足实时跟踪需要，很好的一个DEMO。
camshift tracking algorithm, it will automatically determine the size of target tracking, the center, direction, Operational very fast, real-time tracking needs, a good DEMO. (2007-05-14, Visual C++, 152KB, 下载1298次)
[图形图像处理] kmeans1
k-均值聚类算法实现灰度图像分割，输入图像矩阵和聚类中心个数，返回为最终的聚类中心和图像中每个像素所属类的编号（对应于图像矩阵）
k-means clustering algorithm to achieve gray-scale image segmentation, the input image matrix and the number of cluster centers, the return for the final image of the cluster centers and their respective categories in each pixel number (corresponding to the image matrix) (2008-05-31, matlab, 1KB, 下载952次)
[图形图像处理] ROI
Matlab图像感兴趣区域编码，这个程序所包括的m代码实现了迭代的图像处理，找出感兴趣区域。并同时输出区域、均值、STD、最小值、最大值、中值、中心点等等。
Matlab region of interest image coding, this procedure covered by the code m realize the iterative image processing, to identify areas of interest. And at the same time the output region, the mean, STD, minimum, maximum, median, center and so on. (2008-08-08, matlab, 105KB, 下载754次)
[图形图像处理] isar_code_new
ISAR点目标RD成像算法,距离向包络对齐，多普勒中心相位补偿
Algorithm for ISAR point target imaging (2012-08-02, matlab, 22016KB, 下载622次)
[图形图像处理] checkCir
圆孔测量，使用图像识别技术自动检测圆孔的直径，周长，中心坐标等
hole measurement, the use of image recognition technology automatically detects the hole diameter, circumference, the center coordinates etc. (2007-05-21, Visual C++, 9457KB, 下载588次)
[图形图像处理] fcm
模糊C均值聚类算法处理的图像分割，模糊C均值聚类算法处理的图像分割
An image segmentation method via fuzzy c-means clustering with edge and texture information is proposed. it uses the weighting fuzzy c-mean clustering algorithm to fast implement the segmentation procedure. (2009-10-09, matlab, 2KB, 下载531次)
[图形图像处理] RoadExtraction
高分辨率遥感图像道路提取算法实现，主要思路是利用道路中心点，得到道路中心线。
Road Extraction from High Resolution Images algorithm, the main idea is to use the center of the road, get the road centerline. (2010-09-06, Visual C++, 2954KB, 下载462次)
[图形图像处理] EyesFinder
在AdaBoost检测人脸的基础上，精确找到人脸中眼睛瞳孔中心的程序。
In AdaBoost Face Detection Based on the precision to find human faces in the eye pupil center procedures. (2008-08-17, Visual C++, 1424KB, 下载457次)
[图形图像处理] suijituoyuanjiancesuanfa
介绍了一种椭圆检测的方法，该方法，基于霍夫变换，并根据椭圆几何特征可以快速准确的提取出椭圆中心。本论文为收费文献，现共享，希望对研究计算机视觉以及近景摄影测量的朋友，有所帮助。
Introduce a kind of ellipse detection method, based on the Hough transform, and in accordance with the characteristics of elliptical geometry can be extracted quickly and accurately center oval. This thesis documents for a fee, are sharing the hope that the study of computer vision, as well as close-range photogrammetry friends, be helpful. (2009-03-21, matlab, 301KB, 下载418次)
[图形图像处理] snake
在数字图像中虹膜位置的有效定位是虹膜识别的关键问题。用一种基于主动轮廓线模型的方法定位虹膜的位置，先用灰度投影法检测出瞳孔内的一点作为瞳孔的伪圆心，该圆心只要能落在瞳孔内部即可。然后以该伪圆心为中心，在其周围等角度间隔地取N个点作为初始的snake基准点，按照snake 的运行机制不断进化，直到虹膜的内边界为止。最后，计算进化后的snake形心和snake上的控制点与该形心的距离，取其平均值作为瞳孔的半径，动态轮廓模型的形心作为瞳孔的圆心，即可准确定位出虹膜内边界的位置。实验表明，与常见的定位方法相比，文中的方法速度快、精度高，而且，对瞳孔初始的伪圆心要求不高，鲁棒性更强。
Digital image of the iris in the location of the effective positioning of the key issues of iris recognition. With a model based on the active contour method of positioning the location of the iris, the first gray projection method used to detect the pupil as the point of the pseudo-pupil center, the center of a circle as long as it can fall within the pupil. And then to the pseudo-center as the center, in terms of their spacing around the N points to get the snake as the initial reference point, the operating mechanism in accordance with the snake continued to evolve until the border until the inner iris. Finally, after calculating the evolution of snake and snake-shaped heart of the control point on the centroid distance, choose the average as the radius of the pupil, the dynamic contour model of the centroid as the center of pupil can be an accurate positioning within the borders of the iris position. The experimental results show that the position with common methods, the methods o (2009-04-21, Visual C++, 1723KB, 下载397次)
[图形图像处理] structured-light-feature-extraction
基于Hessian矩阵的条纹中心提取方法提取光条中心点的亚像素图像坐标，并将光条中心点连接成光条中心线 然后利用直线近似算法将光条中心线在拐点处拆分为近似直线:最后将满足角度约束和图像原点到近似直线的距离约束的共线中心点进行融合，对融合后的共线中心点进行直线拟合得到直线光条的参数方程。
Hessian matrix-based extraction method of extracting light stripe center of the center section of the sub-pixel image coordinates, and connected into the center of light bar light bar center line then use the linear approximation algorithm will be the center line of the inflection point of light split into approximately Line: Finally, to meet the angle constraints and images from the origin to the approximate linear constraints collinear fusion center, for a total fusion center for straight line fitting the parameters of linear equation of light stripes. (2011-03-01, Visual C++, 566KB, 下载349次)
[图形图像处理] sliding_windows
本程序首先用Harris角点检测器在输入图像中检测出角点，然后以每个角点为中心，分割出一个15*15大小的滑动窗口，以为后续的相关匹配等做准备。
This procedure first used Harris corner detector in the input image to detect the corner, and then to each corner point as the center, split a 15* 15 the size of sliding window, thinking that the relevant follow-up in preparation for matching. (2008-04-21, matlab, 4KB, 下载349次)
[图形图像处理] MATLABcode
利用Retinex中心环绕算法进行阴影去除，飞常详细并含有程序对应的PPT
Centers around the use of Retinex algorithm to remove the shadow, flying often detailed and contain procedures for the corresponding PPT (2009-10-16, matlab, 17KB, 下载347次)
[图形图像处理] LBP
纹理模式的LBP特征表示及分类原理：就是通过定义一个模板，对于原图像中的每个点，使用该模板进行操作，模板中心点近邻区域对应的图像象素值与中心位置对应的象素值进行比较，且对于不同位置的比较结果赋予不同的权值，从而，变换图像中对应点的值就由模板对应各位置的比较结果之权和表示。
LBP texture pattern feature representation and classification theory: that by defining a template for the original image of each point, use the template to operate, the template center of neighboring regions corresponding to the center of the image pixel values corresponding to the pixel value comparison, and the comparison results for different locations given different weights, and thus, changing the value of the image corresponding points on the template corresponding to each position by the comparison of the results of the right to and express. (2010-01-05, matlab, 2KB, 下载301次)
[图形图像处理] TextureLBP
LBP纹理特征提取算法。首先将检测窗口划分为16×16的小区域（cell），对于每个cell中的一个像素，将其环形邻域内的8个点（也可以是环形邻域多个点，如图 3‑ 4. 应用LBP算法的三个邻域示例所示）进行顺时针或逆时针的比较，如果中心像素值比该邻点大，则将邻点赋值为1，否则赋值为0，这样每个点都会获得一个8位二进制数（通常转换为十进制数）。然后计算每个cell的直方图，即每个数字（假定是十进制数）出现的频率（也就是一个关于每一个像素点是否比邻域内点大的一个二进制序列进行统计），然后对该直方图进行归一化处理。最后将得到的每个cell的统计直方图进行连接，就得到了整幅图的LBP纹理特征
LBP texture feature extraction algorithm. First detection window is divided into 1616 small area (cell), one pixel in each cell, its circular neighborhood of 8 points (You can also ring neighborhood, as shown in Figure 3-4 example of application of the LBP algorithm, three neighborhood shown) clockwise or counterclockwise, the center pixel value than the adjacent points, then adjacent points assigned to 1, otherwise the assignment is 0, so that each point will an 8-bit binary number (usually converted to a decimal number). And then computing the histogram of each cell, each number (assuming that is a decimal number) the frequency (that is, one on each pixel next to the interior point for a binary sequence of statistics), then the histogram normalized. Finally, get the histogram of each cell to connect, to get the whole image of the LBP texture features (2012-06-26, Visual C++, 4KB, 下载298次)
[图形图像处理] ssr_retinex_grahics
中心/ 环绕Retinex 算法。自从Retinex 理论提出以来, Land 本人和其他
人先后提出了Retinex 不同的变种, 最新的一个版
本是中心/ 环绕Retinex 7 。该方法比较以前的版
本, 运算速度显著提高, 而且物理意义更加清楚。
这种运算与人类视网膜和大脑皮层神经元的神经生
理学函数相关, 其形式类似于自然视觉科学中常用
来模拟单个神经元感受野的高斯差分函数
center/around Retinex algorithm. Since Retinex theory, Land I and others have raised Retinex different varieties, The latest version is a center/around Retinex 7. The method compared with the old version, computational speed significantly improved, but more clear physical meaning. Such operations and human retinal neurons in the brain and nerve function related physiology, form similar to the natural visual science used to simulate the feeling of individual neurons in the wild Difference of Gaussian function (2007-04-14, matlab, 4KB, 下载298次)
[图形图像处理] leituoyuantezhengzidongshibiejiyaxiangsutiqudewanz
针对视觉检测中高精度、自动化类椭圆图像特征提取的需要,提出一种类椭圆特征自动识别和中心亚像
素定位的完整实现方法。方法中采用基于类椭圆边缘属性对特征区域进行自动识别,采用最小二乘椭圆拟合
精确求取类椭圆亚像素定位中心。实验表明,基于该解决方案,可以实现类椭圆特征的自动识别及椭圆中心的
亚像素定位,定位算法精度较高,鲁棒性强,同时简便易行,不需要人机交互,可以很好地满足视觉检测仪器化
需求。
Vision testing for high-precision, automated image feature extraction Elliptic needs, the characteristics of a Class of Elliptic Automatic Identification and central sub-pixel positioning methods full implementation. Methods based on the type of property on the edge of the characteristics of an oval region of automatic identification, using least squares ellipse fitting precision to strike a class of elliptic center sub-pixel positioning. The experimental results show that based on the solution, you can realize the characteristics of Class of Elliptic Automatic Identification and oval center sub-pixel positioning, high precision positioning algorithm robustness, and user-friendly and does not require human-computer interaction can be well Vision testing instruments to meet the needs. (2009-03-21, matlab, 177KB, 下载293次)
[图形图像处理] MeanShift
meanshift聚类，找到聚类中心,以及数据的类别
Clustering meanshift to find cluster centers, as well as the type of data (2007-11-07, matlab, 4KB, 下载293次)
[图形图像处理] chirplet_analasis
一种自适应chirplet分解的快速算法
针对信号自适应chirplet分解未知参数多、实现起来比较困难的特点，文献[1]提出了一种新的chirplet分解快速算法。该算法利用计算信号的二次相位函数，得到其能量分布集中于信号的调频率曲线上的结论，此时通过谱峰检测可同时获得chirplet调频率、时间中心和幅度的估计；然后通过解线性调频技术获得其初始频率和宽度的估计，仿真结果验证了本文算法的有效性。
Chirplet an adaptive decomposition algorithm for rapid adaptive signal decomposition Chirplet many unknown parameters, is more difficult to achieve the characteristics of the literature [1] proposed a new fast algorithm Chirplet decomposition. The algorithm calculating the signal function of the second phase, its energy distribution to be concentrated in the signal modulation frequency curve on the conclusions, this time through the spectral peak detection can be obtained simultaneously Chirplet tune the frequency, timing and magnitude of the estimated center then through the solution line technology of FM frequency and the width of the initial estimates, the simulation results verify the effectiveness of the algorithm. (2009-03-09, matlab, 79KB, 下载277次)
[图形图像处理] findcenter
用于求取灰度图光斑的中心，可以用于寻找光斑图像的中心，如激光光斑，
Used to strike a grayscale facula center spot can be used to find the center of the image, such as the laser spot, (2007-09-15, matlab, 126KB, 下载274次)