基于Gabor变换的多角度人脸表情识别方法  

Multi-Angle Facial Expression Recognition Method Based on Gabor Transform

作　　者：王康毅 邵苏杰 WANG Kang-yi;SHAO Su-jie(Department of Computer Science,Changzhi University,Changzhi Shanxi 046011,China;School of Computer Science,Beijing University of Posts and Telecommunications,Beijing 100876,China)

机构地区：[1]长治学院计算机系,山西长治046011 [2]北京邮电大学计算机学院,北京100876

出　　处：《计算机仿真》2024年第4期233-236,526,共5页Computer Simulation

基　　金：山西省高校科技创新项目(2019L0915)。

摘　　要：由于人脸外形的不稳定性,可通过人脸变化产生多种表情,在不同的观察角度上人脸视觉图像存在较大差异。且在光照变化、面部表情姿态以及遮挡等因素的影响下,难以准确提取人脸表情特征,导致识别准确率偏低。为此,提出基于Gabor变换的多角度人脸表情识别方法。通过人眼定位,对多角度人脸表情图像完成几何预处理,提升人脸表情识别精度。采用Gabor变换方法提取多角度人脸表情图像特征。利用弹性模板匹配方法对特征关键点开展弹性网格匹配,计算出图像的代价函数。采用K-近邻分类策略匹配评估多角度人脸表情图像,完成多角度人脸表情识别。实验结果表明,以上方法的识别时间在2s内,识别准确率接近100%,应用性能优于已有方法,验证了研究方法有效性更强、精准性更高。Due to the instability of the face shape,there are significant differences in the visual images of faces from different observation angles.Under the influence of illumination change,facial expression posture and occlusion,it is difficult to accurately extract facial expression features,leading to low recognition accuracy.Therefore,a multiangle facial expression recognition method based on Gabor transform was proposed.Through the eye location,the geometric preprocessing of multi-angle facial expression image was completed,so that the accuracy of facial expression recognition could be improved.Gabor transform was used to extract the feature of the multi-angle facial expression image.Moreover,the elastic template matching method was adopted to perform the elastic mesh matching on the feature key points,and thus to calculate the cost function of the image.Finally,the K-nearest neighbor classification strategy was used to match and evaluate multi-angle facial expression images,thus completing the multi-angle facial expression recognition.Experimental results prove that the recognition time of the proposed method is less than 2s,and its recognition accuracy is close to 100%,so the method is more effective and accurate.

关 键 词：图像特征提取 图像预处理 弹性模板匹配 近邻分类策略 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]