搜索资源列表
ssim_index.rar
- 该程序SSIM(结构相似度)是基于matlab平台,用于图像的客观质量评价,是FR中精典算法,The program SSIM (structural similarity) is based on matlab platform for an objective image quality evaluation, is in the classical algorithm FR
image similarity
- 图像相似度检测,下面是直方图相交的代码,同种图片的识别率达90%以上,性能非常稳定。 程序的例子是8位(256色)位图,其他位图类似。 -Image similarity detection, the following is the histogram intersection of the code, the same kinds of picture identification rate of 90, performance is very stable. Procedure
FastAlgorithmofImageMatching
- Wavelet t ransform and projection were referred to after wavelet t ransform projection and se2 quential similarity detection algorithm(SSDA) were applied to the low f requency part of the image to get a set of potential matching point-Wavelet t
imagematching
- 数字图像匹配matlab源码,就是指图像之间的比较、得到不同图像之间的相似度。基于数字图像,编写对两副数字图像进行匹配的算法及演示程序。-Digital image matching matlab source, refers to a comparison between images, the similarity between different images. Based on digital images, the preparation of two digital image m
pic
- 实现BMP图像的直方图均衡化,并集成了两幅图片的相似度匹配,能输出匹配度-Achieve BMP image histogram equalization, and integrated picture of the similarity of the two match, output match degree
cbir
- 用的是局部颜色特征,再说细点是用里面的区域颜色直方图的方法。把图像归一化到256X256,把图像分成4X4块,计算16个区域的颜色直方图、、、 最后计算相似度是用欧氏距离.-Using local color feature, repeat fine-point is inside the regional color histogram method. The normalized image to 256X256, the image is divided into 4X4 blocks
biye
- 基于投票算法的目标跟踪,基于二阶非线性投票的多目标跟踪算法。该算法通过目标匹配得到同一目标在不同帧中的位置,同时利用特征监测来处理目标的遮挡、分裂问题,并实现目标特征的实时更新。在目标匹配过程中,通过对目标前一帧与当前帧的特征相似性进行投票,得到匹配目标。利用视频图像进行实验,结果表明:该方法对噪声、阴影、遮挡、分裂等具有良好的鲁棒性,较好地实现了多目标的跟踪。-The method used object matching to get objects’ position in differe
GPAC
- 多区域的分割方法,用到分水岭及水平集等多种算法-A Variational Framework for Multi-Region Pairwise Similarity-based Image Segmentation
ssim_index
- This an implementation of the algorithm for calculating the Structural SIMilarity (SSIM) index between two images.- This is an implementation of the algorithm for calculating the Structural SIMilarity (SSIM) index between two images.
An-Efficient-Method-of-Texture-Synthesis-Based-on-
- 提出一种带边界匹配的基于 Graph Cut 的快速纹理合成算法.通过将纹理样本以不同的位移贴到输出图中完成合成 ,重叠区域的像素取值由 Graph Cut 确定.引入边界图辅助位移搜索 ,以增强合成结果的边界连续性 在预处理过程中计算 2 个相同样本在所有相对位移下的匹配误差 ,选取一部分误差最小的位移组成 “优选位移” 集合 ,合成过程中的块间相对位移仅从此集合中选取 ,大大地提高了合成速度.-This paper proposes an efficient method for tex
match
- 由图像的颜色直方图求取相应的区域颜色熵,利用该信息判断出两幅图像间的相似度,可将其利用到诸如镜头检测之类的场合中-Use color histogram to calculate corresponding regional color entropy, use this information to determine the similarity between two images,which can be used to detect the occasions such as shot
similarity
- 自相似特征(self-similarity)描述子的提取代码,算法见“Matching Local Self-Similarities across Images and Videos”-Self-similar characteristics (self-similarity) describe the extraction of sub-code, algorithms, see " Matching Local Self-Similarities across Images and
New_Dominant_Color
- Calculates dominante color and measure the similarity between the input image and directory contains other images
ssim-1.1
- Implementation of the Self Similarity descr iptor based on the paper: Matching Local Self-Similarities across Images and Videos, Eli Shechtman and Michal Irani CVPR 07-Implementation of the Self Similarity descr iptor based on the paper: Matchi
Interactive_Image_Segmentation_by_Maximal_Similari
- 实现了区域合并的图像分割的经典方法,是Interactive Image Segmentation by Maximal Similarity based Region Merging 论文的源码,可以好好研究-Regional integration to achieve the classic method of image segmentation is the Interactive Image Segmentation by Maximal Similarity based Regio
shibie
- 基于奇异值分解的人脸识别方法 梁毅雄 龚卫国 潘英俊 李伟红 刘嘉敏 张红梅 提出了一种将傅里叶变换和奇异值分解相结合的人脸自动识别方法.首先对人脸图像进行傅里叶变换,得到其具有位移不变特性的振幅谱表征.其次,从所有训练图像样本的振幅谱表征中给定标准脸并对其进行奇异值分解,求出标准特征矩阵,再将人脸的振幅谱表征投影到标准特征矩阵后得到的投影系数作为该人脸的模式特征.然后,对经典的最近邻分类器算法进行了改进,并采用模式特征之间的欧式距离作为相似性度量,从而完成对未知人脸的识别.采用ORL
SSDA
- SSDA序贯相似性检测方法对图像进行模板匹配。从源图像中取小图,再到源图像中找到小图所处的位置。-SSDA sequential similarity detection method of the image template matching. Source image taken from the small map, to find a small map the source image position.
SimilarityCompare
- 本程序用于计算两个矢量图形的相似程度。 对于图形形状相似性问题,本程序满足了五个基本要求: 1.对于任意两个图形的相似程度必须得出一个量化的结果,在此称为图形相似度。 2.对图形形状的检测必须忽略 大小、旋转、轴对称、连线顺序的影响。 3.对于相同的图形,形状相似度必须为1;对于不相同的图形,形状相似度必须小于1。 4.两个图形的形状相似度必须与其相似程度成严格单调性,即对于同一个基准图形,越相似的图形相似度越高,越不相似的图形相似度越低。 5.必须能在可接受的时间与空间
HSV-color-space-based-on-similarity-calculation-me
- 一种基于HSV空间的颜色相似度计算方法 一种基于HSV空间的颜色相似度计算方法-HSV color space based on similarity calculation method
similarity
- python写的图像相似度算法。比较图像相似度,选出最相思的图片-python write the image similarity algorithm. Compare image similarity, image to select the most Acacia