Description: 本文的题目是基于分形和遗传算法的人脸识别方法，对有限人群提出一种采用分形特征和遗传聚类的识别方法: 将图像分成很多小区域, 分别计算各个区域的分形特征, 以充分利用图像二维信息 同一个模式有多个样本, 通过遗传算法进行聚类以得到最优解实现不变性识别. 最后采用ORL 人脸图像库的一组图像对比了新方法、本征脸法和自联想神经网络方法, 结果表明该方法的识别率, 与本征脸法相似, 比自联想神经网络高.-The title of this article is based on fractal and genetic algorithms for face recognition method, a crowd of limited use of fractal characteristics and the identification of genetic clustering methods: the image is divided into many small regions, each region were calculated fractal characteristics, to take full advantage of two-dimensional image information with a model for a number of samples, through the genetic clustering algorithm in order to obtain the optimal solution to achieve invariant recognition. Finally, using ORL face image database of a group of image contrast of the new methods, eigenface law and auto-associative neural network methods, results show that the method of recognition rate, with the eigenface method is similar to auto-associative neural network than high. Platform: |
Size: 380928 |
Author:阳关 |
Hits:
Description: 基于肤色分割的人脸检测，有肤色分割，排除假区域，特征验证
-Color segmentation based on face detection, color segmentation there, removal leave the region, feature verification Platform: |
Size: 1024 |
Author:zhaowenbin |
Hits:
Description: 系统有以下部分组成：电脑自带摄像头拍照、人脸检测、将人脸照片录入数据库、输入照片进行人脸识别。本程序是基于肤色识别的方法对人脸进行检测，人脸肤色范围是100≤B≤120，140≤R≤160，所以将此范围内的像素点置白，剩余部分置黑。利用imerode函数对图片进行球状腐蚀，然后再对图片进行中值滤波，达到平滑效果。最后，对于这张已经缩放而且二值化和各种处理之后的照片来说，如果白化区域的像素点少于1000，就舍弃。将图片进行分割，这里我们引进了欧拉数。
这样就可以把一些类似颜色人脸的背景排除。最后对于每块白化区域，确定其边界然后画出，便可以完成多人脸检测。
-System has the following components: The computer comes with the camera, camera, face detection, face recognition face photo entry database, enter the photo. This procedure is a method based on color recognition, face detection, face color range 100 ≤ B ≤ 120,140 ≤ R ≤ 160, so set white pixels within this range, the remaining part of the set black. Globular corrosion imerode function of the picture, then picture the median filtering to achieve a smoothing effect. Finally, for this has been scaled but binarization and various processing after pictures, if the pixel point of the whitened area is less than 1000, it is discarded. Image segmentation, we introduce the Euler number. This can exclude the background of similar color face. Finally, for each block of albino region and then draw its boundary, we can complete multiple face detection. Platform: |
Size: 301056 |
Author:魏宗帅 |
Hits:
Description: 基于C#编写的人脸识别系统，首先进行人脸图片的预处理，包括：标记人脸区域，光照补偿，灰度化，高斯平滑，直方图均衡化处理，对比度增强，二值化变换，去除孤立点等操作，然后标记人脸特征点，提取特征信息，与数据库中信息比较进行识别-Written in C#-based face recognition system, the first face image preprocessing, including: labeled human face region, illumination compensation, graying, Gaussian smoothing, histogram equalization, contrast enhancement, binarization transformation to removeisolated point operation, and then mark the facial feature points extracted feature information compared with the information in the database to identify Platform: |
Size: 5063680 |
Author:王华伟 |
Hits:
Search - face Database region