CAJ | 학술논문

针对唇读识别中唇形状态的描述问题，提出一种基于 Fourier 描述子的唇形分类方法。该方法通过迭代演算获得嘴唇的位置与大小，先通过边缘侦测法找出唇形边缘；再由 Fourier描述子找出唇形外形的重要特征值，进而表达唇形状态的有效信息；最后将转换出的 Fourier描述子经过正规化处理后，输入到人工神经元网络中进行分类。实验结果表明，该方法的唇形分类正确率较高，且耗时较少，为唇读识别中各状态的确定提供了依据。
침대진독식별중진형상태적묘술문제，제출일충기우 Fourier 묘술자적진형분류방법。해방법통과질대연산획득취진적위치여대소，선통과변연정측법조출진형변연；재유 Fourier묘술자조출진형외형적중요특정치，진이표체진형상태적유효신식；최후장전환출적 Fourier묘술자경과정규화처리후，수입도인공신경원망락중진행분류。실험결과표명，해방법적진형분류정학솔교고，차모시교소，위진독식별중각상태적학정제공료의거。
Aiming at the describing the problem of lip reading status,the authors presented a method to classify lip shapes based on Fourier descriptor.Localization of the lip region can be found by an iterative calculation method.Firstly,Canny’s edge detection method was used in order to define the lip contour.Then,in order to find the best parameters of lip’s shape,a Fourier descriptor was used. Important parameters allowing us to express the useful information of lip’s shape status were then extracted.Finally, the different parameters characterizing the input lip were classified into lip categories. The experimental results show that this method has a higher accuracy of lip’s classification,consumes less time,and provides a clear basis for each status during lip recognition.