CAJ | 학술논문

为实现非闭合曲线的精确识别，基于傅里叶变换和几何对称性，提出了一种用于识别非闭合曲线的新算法。首先，将非闭合曲线二值化，并生成表示此非闭合曲线的点集S1；然后，根据几何对称性生成新的点集S2，将点集S1和S2的首尾点相连，形成由S1和S2表示的封闭二值曲线；最后对新生成的封闭曲线进行傅里叶描述，生成描述此封闭曲线的傅里叶描述符。根据傅里叶变换原理，对傅里叶描述符进行归一化处理。同时，引入χ2置信度来衡量曲线的相似度。实验结果表明：所提出的傅里叶描述符可以实现非闭合曲线的精确识别和重构，且不失旋转、尺度及平移不变性；与传统的傅里叶描述符相比，该傅里叶描述符具有更低的置信度和更好的数据稳定度。
위실현비폐합곡선적정학식별，기우부리협변환화궤하대칭성，제출료일충용우식별비폐합곡선적신산법。수선，장비폐합곡선이치화，병생성표시차비폐합곡선적점집S1；연후，근거궤하대칭성생성신적점집S2，장점집S1화S2적수미점상련，형성유S1화S2표시적봉폐이치곡선；최후대신생성적봉폐곡선진행부리협묘술，생성묘술차봉폐곡선적부리협묘술부。근거부리협변환원리，대부리협묘술부진행귀일화처리。동시，인입χ2치신도래형량곡선적상사도。실험결과표명：소제출적부리협묘술부가이실현비폐합곡선적정학식별화중구，차불실선전、척도급평이불변성；여전통적부리협묘술부상비，해부리협묘술부구유경저적치신도화경호적수거은정도。
In order to accomplish accurate recognition of non-closed curves,a new Fourier descrip-tion of non-closed curves is proposed based on Fourier transform and geometric symmetry.First,the binarization of the non-closed curve is carried out and a set of points S1 is generated.Then,accord-ing to geometric symmetry,a new set of points S2 is generated.The start point of S1 is connected with the end point of S2 ,and a closed binary curve described with S1 and S2 is formed.Finally,Fou-rier description is applied to the new closed contour,and the Fourier descriptors are obtained to re-present the new closed contour.Based on the principle of Fourier transform,Fourier descriptors are normalized.Meanwhile,the χ2 confidence level is introduced to measure the similarity of curves. The experimental results show that the proposed Fourier descriptor can be used to accurately distin-guish and reconstruct a non-closed curve,and is invariant under rotation,scaling and translation for non-closed curves.Compared with traditional Fourier descriptors,the proposed Fourier descriptor has lower confidence level and better data stability.