摘要
提出了一种融合轮廓和区域信息的新的形状描述子。首先提取图像的轮廓,并对轮廓进行采样,形成描述形状轮廓的有序点列表。之后对每个采样点分别顺时针和逆时针等距离跟踪获得两个邻点,计算相邻点距离、拱高以及质心距离三个函数的均值、方差和极差。然后获取由拱高和质心距离组成的复函数的傅里叶描述子。组合统计值与傅里叶描述子形成轮廓特征。最后对区域进行采样形成形状矩阵,获得低频傅里叶系数作为区域特征。对MPEG—7标准图形库的检索实验显示,该描述子的检索性能显著优于三角形面积函数、质心距离函数、拱高半径复函数和Hu不变矩等同类描述子。
A new shape descriptor with fusion contour and region information is proposed. Firstly the contours of image are extracted and sampled, and the ordered list of points are gotlen. Then, clockwise and counterclockwise track each sampling point equidistant distance to get two adjacent points, and the distance of adjacent points, arch- height and centroid distance are calculated. After that, their mean, variance and range are counted. Then a com- plex function of arch-height and eentroid distance are and eonstrucled the Fourier coefficients are obtained. A con- tour feature is formed which is composed of Statistics and Fourier descriptors. Finally, the shape matrices are got through area sampling, and use the low-frequency Fourier coefficients as regional features. Retrieval experiments on MPEG--7 shape database show that, the retrieval performance of this descriptor is better than the triangle area func- tion, the centroid distance function, arch-height radius complex function, and Hu invariant moment significantly.
出处
《科学技术与工程》
北大核心
2012年第35期9570-9575,共6页
Science Technology and Engineering
基金
宿迁学院科研基金(2011KY21)资助
关键词
区域特征
形状特征
图像检索
傅里叶描述子
region-based features shape features image retrieval Fourier descriptors
