人脸表情识别技术研究
人脸表情识别技术研究(任务书,开题报告,论文13000字)
摘 要
随着计算机技术的发展,智能化是计算机研究的主题,而人脸表情识别技术作为当前热门的计算机视觉的一个重要研究方向,其在社会的各个领域均有可预见的巨大发展潜力。
本文研究了对静态图像的表情识别,研究步骤主要分为图像的预处理,表情特征提取,表情识别等,研究内容如下:
1.检索到表情特征库,本文采用了日本研究机构的JAFFE人脸表情数据库,分析了此数据集的特点与优势。
2.进行图像的读取与预处理,在使用人脸表情数据库的基础上进行图像的读取与预处理,主要包括图像格式的转换,灰度化,使用高通滤波器以及低通滤波器对图像进行处理,边缘检测等。
3.研究了传统的特征提取算法,并主要研究了主成分分析方法以及LBP方法的数学原理以及其算法实现,并比较了它们的优缺点。
4.对谷歌神经网络库tensorflow进行了分析,并在此基础上实现了基于cnn的人脸表情识别系统
关键词:人脸识别;特征提取;卷积神经网络
Abstract
This With the development of computer technology, intelligence is the subject of computer research, and face expression recognition technology as an important research direction of the current hot computer vision, it has a foreseeable great potential for development in all fields of society.
[资料来源:http://doc163.com]
In this paper, the expression recognition of static images is studied, and the research steps are mainly divided into image preprocessing, expression feature extraction, face recognition, expression recognition, etc., the research contents are as follows:
1. In this paper, the JAFFE facial expression database of Japanese research institutions is used, and the characteristics and advantages of this dataset are analyzed.
2. Image reading and preprocessing, in the use of face expression database on the basis of image reading and preprocessing, mainly including image format conversion, grayscale, the use of high-pass filter and low-pass filter image processing, edge detection, image cropping and rotation.
3. The traditional feature extraction algorithms are studied, and the principal component analysis (PCA) method and the mathematical principle of LBP method are studied, and their advantages and disadvantages are compared.
4.This paper introduces the tensorflow of Google Neural Network library, and studies the expression recognition method based on deep learning. [资料来源:www.doc163.com]
Key Words:Face recognition;Feature Extraction;Convolutional Neural Networks
目录
第1章绪论 1
1.1研究现状 1
1.2人脸识别与表情识别 2
1.3本文主要内容 2
1.4小结 3
第2章图像预处理 4
2.1灰度化 4
2.2高通/低通滤波器 5
2.3图像二值化 5
2.4边缘处理 6
2.5小结 8
第3章静态表情特征提取方法 9
3.1表情特征算法综述 9
3.2PCA主成分分析方法 10
3.3LBP局部二值模式 14
3.4PCA与LBP的比较 17
第4章人脸表情识别系统实现 18
4.1人脸表情数据库的选择与分析 18
4.2系统技术实现 19
4.3 CNN卷积神经网络 19 [资料来源:Doc163.com]
4.3.1向前传播阶段 20
4.3.2向后传播阶段 23
4.4训练参数设置 23
4.2系统实现过程及结果分析 24
第5章总结与展望 30
5.1总结 30
5.2展望 30
参考文献 31
致谢 32 [资料来源:http://Doc163.com]