基于RGB-D图像的手语识别算法研究

手语作为表达信息的一种方式，是聋哑人与外界交流的最主要途径。随着人工智能技术的发展，新型、自然、便捷的人机交互方式已经成为当今各行业交互方式的新趋势。传统手语识别方法普遍存在背景干扰严重、特征提取不完整及识别准确率低的缺点。针对以上问题进行研究，提出基于彩色-深度（RGB-D）图像的手语识别算法。在静态孤立词手语识别的基础上测试，实验对比结果表明，基于RGB-D的方法能得到比传统手语识别算法更好地性能；利用深度信息处理图像，有效地去除了场景中光照、背景的干扰，同时结合肤色信息去除非手形部分；采用概率神经网络（PNN）手语分类器对孤立词的识别准确率超过90%，取得了较高的识别率。

Research of Sign language recognition algorithm based on RGB-D image

Sign language as a way to express information, it is the most important way for the deaf communicate with the outside world. With the development of Artificial intelligence technology, the new, natural and more convenient way of human-computer interaction has become the new trend of interaction in various industries today. The traditional sign language recognition method has the disadvantage of serious background interference, incomplete feature extraction and low recognition accuracy. A speech recognition algorithm based on color and depth (RGB-D) image is proposed to solve the above problems. The experimental results show that the RGB-D based method can obtain better performance than the traditional sign language recognition algorithm. Using depth information to deal with images, effectively remove the interference of light and background in the scene, and combine the color information to the hand. Using the probability neural network (PNN) sign classifier, the recognition accuracy of the isolated words was over 90%. and it achieved a higher recognition accuracy rate.