需要金幣： 1000 個金幣 |
資料包括：完整論文 |  |
 |
| 轉(zhuǎn)換比率：金額 X 10=金幣數(shù)量， 例100元=1000金幣 | 論文字?jǐn)?shù)：18641 |  | |
| 折扣與優(yōu)惠：團(tuán)購最低可5折優(yōu)惠 - 了解詳情 | 論文格式：Word格式(*.doc) |  |
摘要:隨著人工智能理論與技術(shù)的發(fā)展，計算機(jī)視覺技術(shù)取得了長足的進(jìn)步，已經(jīng)漸漸成為國防建設(shè)、公共安全監(jiān)控、醫(yī)療衛(wèi)生領(lǐng)域應(yīng)用較為廣泛的一項(xiàng)技術(shù)。人臉檢測與識別作為計算機(jī)視覺重要的分支，近年來成為學(xué)術(shù)界研究的重點(diǎn)，同時也有一些應(yīng)用項(xiàng)目的開發(fā)。 本文研究和實(shí)現(xiàn)了AdaBoost人臉檢測算法和基于特征臉的人臉識別算法。結(jié)合AdaBoost檢測器級聯(lián)的特性，引入了基于梯度方向金字塔和支持向量機(jī)的人臉分類算法，去除了部分誤檢測的結(jié)果；引入了基于膚色模型的分割算法，對人臉區(qū)域進(jìn)行了精確定位。實(shí)驗(yàn)在第二代身份證、Feret人臉數(shù)據(jù)庫以及互聯(lián)網(wǎng)圖片資源上進(jìn)行，給出了實(shí)驗(yàn)結(jié)果與分析。算法對正面人臉效果顯著，但對于旋轉(zhuǎn)和傾斜的人臉，魯棒性不強(qiáng)。本文最后進(jìn)行了總結(jié)，同時提出了后續(xù)研究的方向。 本文最終的實(shí)現(xiàn)在Windows平臺下，利用Visual Studio 2008和OpenCV庫實(shí)現(xiàn)。 關(guān)鍵字：人臉檢測；人臉識別；AdaBoost；梯度方向金字塔（GOP）；支持向量機(jī)（SVM）；特征臉
Abstract:With the development of theory and technology of artificial intelligence, technologies in computer vision had made considerable progress these years, and were widely applied in the fields of homeland security、video surveillance and medical health. Being an important part of computer vision, the theory and technology of human face detection and recognition had draw great attention in academia world. This paper studies and implements the face detection algorithm AdaBoost and face recognition algorithm based on eigenface. Considering the cascade property of AdaBoost algorithm, this paper improves the AdaBoost with a human face classification algorithm based on gradient orientation pyramid and support vector machine. Also an algorithm of image segmentation based on skin color model is introduced to locate the human face accurately. Experiments were carried out on the test set consists of ID-card、Feret database and images from the internet. Experimental results show that both the detection method and recognition method achieve good results on the frontal face images, but less robust on the images of face with apparent rotation and tilt. At last this paper puts forward the future work. The implementation and experiment were developed with Visual Studio 2008 on Windows, and OpenCV library were used. Keywords: Face Detection; Face Recognition; AdaBoost; Gradient Orientation Pyramid; Support Vector Machine; Eigenface
本文實(shí)現(xiàn)了AdaBoost人臉檢測算法，并在這個基礎(chǔ)上實(shí)現(xiàn)了基于特征臉的人臉識別，對于目前的社會網(wǎng)絡(luò)監(jiān)控具有一定的實(shí)際意義。主要的應(yīng)用場景是公共場所視頻的監(jiān)控。對于公安系統(tǒng)對人員的排查和尋找有輔助作用。例如，對網(wǎng)吧、銀行、酒店等公共場所的柜臺、收銀臺進(jìn)行視頻監(jiān)控，可以結(jié)合公安系統(tǒng)的逃犯數(shù)據(jù)庫，對公共場所實(shí)行視頻監(jiān)控，一旦發(fā)現(xiàn)嫌疑人出現(xiàn)在視頻中，經(jīng)過人工確認(rèn)后，實(shí)行抓捕。此外，也可以用于社會人員人連數(shù)據(jù)庫的采集等方面。
|
