摘要
目前一些相对成熟的手势识别算法,如基于模板匹配的方法、基于人工神经网络的方法以及基于隐马尔科夫模型的方法,都存在计算复杂的劣势,而基于深度学习的手势识别具有一定优势。通过深度学习提取多层网络简化的高价值易用特征,通过向量等表示,简化算法以实现良好的识别效果。通过摄像头采集室内复杂背景下的手势图像,在计算能力、存储能力强大的PC平台通过深度学习处理图像,提取特征,然后进行分类识别,能提高识别准确率。通过改进硬件或算法还可提高识别效率及安全性。
At present,there are some relatively mature recognition algorithms of hand gestures based on template matching method,artificial neural network method or hidden Markov model.These methods has the disadvantages of computational complexity.There fore,some experts and scholars have studied gesture recognition based on deep learning,extracting high-value and easy-to-use fea tures simplified by multi-layer network through deep learning,thus simplifying the algorithm to achieve excellent recognition effect.Eventually it is concluded that hand gesture recognition based on deep learning have advantages is improved.The natural images of hand gestures in complex indoor background are captured by camera.The images are processed to extract features through deep learn ing and then classified by other classification algorithms on the PC platform with extremely good computing and storage ability,which can improve the accuracy of the final recognition.The efficiency and security also can be improved by improving the algorithm or hard ware.
作者
沈雅婷
SHEN Ya-ting(School of Computer Science,Nanjing University of Science and Technology Zijin College,Nanjing 210046,China)
出处
《软件导刊》
2019年第11期25-29,共5页
Software Guide
关键词
深度图像
预处理
特征提取
深度学习
手势识别
deep image
preprocessing
feature extraction
deep learning
gesture recognition
