基于Sugeno模糊模型的ANFIS在手势识别中的应用  

The application of ANFIS in gesture recognition based on Sugeno fuzzy model

作　　者：蒋磊磊[1] 秦丽娟[1] 李武举 

机构地区：[1]沈阳理工大学信息科学与工程学院,辽宁沈阳110159

出　　处：《微型机与应用》2016年第24期50-53,共4页Microcomputer & Its Applications

摘　　要：手势识别正在成为人机交互技术研究中的一种重要模式,运用摄像机和计算机视觉技术捕获静态手势图像,在YCr Cb空间下利用高斯分布对肤色建模,通过计算背景颜色像素概率分割出干净的手势图像,再采用基于Sugeno模糊模型的自适应神经-模糊推理系统识别手势。与传统的神经网络识别系统相比,该识别方法具有较好的自适应性和鲁棒性。实验结果表明,基于Sugeno模糊模型的自适应神经-模糊推理系统能够有效地进行特征学习,正确识别率可以达到95%左右,是一个非常有效的静态手势识别方法。Gesture recognition is becoming an important mode in human-computer interaction technology research. It uses cameras and computer vision technology to capture static gesture image, and in YCrCb color space, uses the Gaussian to distribut skin color modeling. By calculating the probability of background pixel color, clean gesture image is segmented. Then the Sugeno fuzzy model based adaptive neuro fuzzy inference system is used to get gesture recognition. Compared with the traditional neural network recognition system, the recognition method has better a- daptive and robustness. The experimental results show that based on Sugeno fuzzy model and adaptive neural fuzzy inference system can effec- tively learn the feature, the correct recognition rate can reach 95% or so. It is a very effective method of static gesture recognition.

关 键 词：人机交互 手势 基于Sugeno模糊模型 自适应神经-模糊推理 

分 类 号：TP751[自动化与计算机技术—检测技术与自动化装置] TP753[自动化与计算机技术—控制科学与工程]