基于视觉的手势识别算法及应用的研究

Research on gesture recognition algorithm and application based on vision

随着计算机快速的发展以及广泛的应用,人机交互技术已成为人们日常生活中不可或缺的部分,而手势识别则顺应了发展需要。原始手势图像需要做一些预处理操作才能更好的进行下一步操作分析。本文主要针对基于视觉的手势图像设计了 Alex CNN卷积神经网络模型,研究了图像与处理算法,包括数据增广、肤色提取、形态学处理等操作对所设计的神经网络模型准确率进行测试。结果表明用经过图像预处理后的图像所训练得到的模型准确率相比用原始图像进行训练所得到的模型准确率大为提升。

With the wide application of computer,human-computer interaction has become an important part of people's daily life,and gesture recognition meets the development needs.The original gesture image needs to do some preprocessing operations in order to better analyze the next operation.In this paper,Alex CNN is designed for gesture images based on vision Convolution neural network model,studied the image and processing algorithm,including data augmentation,skin color extraction,morphological processing and other operations,the impact on the accuracy of the designed neural network model.The results show that the accuracy of the model trained by the image preprocessing is greatly improved compared with the accuracy of the model trained by the original image.