复杂光照下手势识别技术的研究.docVIP

  • 8
  • 0
  • 约4.9万字
  • 约 63页
  • 2017-02-10 发布于重庆
  • 举报
复杂光照下手势识别技术的研究

摘 要 新世纪以来,随着计算机水平的不断提高,人类日常生活、学习和工作已经越来越离不开计算机了。在计算机的使用过程中,人们对计算机的性能和智能这两方面提出更高的要求。研究出和谐自然的人机交互方式是近年来计算机领域研究的重要课题,而主要内容之一便是手势识别技术。 目前手势识别的难点主要集中于手势分割以及手势识别这方面。在手势分割时,分割的效果受到光照变化的影响,因此,文中提出一种结合颜色均衡和椭圆模型的手势分割新方法来降低光照变换对分割结果的影响;在手势识别时,由于BP神经网络在训练时容易陷入局部极小而得不到全局优化,文中提出一种结合遗传算法和BP神经网络的手势识别新方法,以此来提升识别率。 本文主要对静态手势识别做了比较深入的研究,具体工作如下: (1)主要介绍图像预处理的常用方法。首先对图像的几何变换进行介绍,并实验验证本文所用的三种几何变换。其次介绍图像的平滑,并对中值滤波进行理论分析和实验验证。最后介绍锐化,实验验证Sobel算子的锐化效果。 (2)分析了常用的手势图像分割方法,提出一种基于颜色均衡和椭圆模型的手势分割新方法,并实验验证了本文所提出的手势分割新方法的分割效果。 (3)首先对常见特征参数的优缺点做适当的分析,然后介绍Hu矩的概念、性质及定义。计算手势的Hu矩,通过分析数据,验证手势Hu矩的不变性和可分性。 (4)提出一种基于遗传算法和BP神经网络的手势识别方法,并实验证明了本文所提出的手势识别方法的识别效果。对比试验表明,本文的分类算法可以克服训练时陷入局部极小的困扰,明显提高识别率。 关键词:手势分割;手势识别;肤色椭圆模型;BP神经网络;遗传算法 Abstract In recent years, with the continuous improvement of the level of computer, human daily life, learning and work have become more and more can not be separated from the computer. In the course of the use of the computer, people put forward higher requirements on the performance and intelligence of the computer. The need for human development has not been done in the traditional way. The study of human computer interaction is an important topic in computer field in recent years, and the main content of the research is hand gesture recognition technology. At present the difficulty of gesture recognition mainly focused on the division of gestures and gesture recognition. When gesture segmentation, segmentation effect is affected by illumination change, therefore, a combination of color balance and elliptical model is proposed in this paper, a new method of gesture segmentation to reduce the effects of light transform to the segmentation result; In gesture recognition, because while training the BP neural network easy to fall into local minimum and can not get global optimization, this paper proposes a combination of genetic algorithm and BP neural network, a new method of gesture recognition, in order to improve the recognition rate. The writer studied the key algorithm pr

文档评论(0)

1亿VIP精品文档

相关文档