一种大数据背景下的图像特征提取算法

An Image Feature Extraction Algorithm Based on Big Data

提出一种基于中轴变换的图像骨架特征提取算法,结合Hausdoff距离方法与二次误差度量方案,对图像数据进行简化,最终获得简洁鲁棒的骨架特征。使用自定义的数据集进行验证,实验证明,使用中轴变换的图像骨架提取方法对边界噪声具有良好的抵抗性,在不明显影响提取骨架速度的同时,能够保证图像骨架的精度。

An image feature extraction algorithm of image contour based on medial axis transform was proposed in this paper.Combined with Hausdoff distance method and quadric error metric,the image data can be simplified to get the concise robust contour feature.Experiments demonstrate by the custom data set that,on one hand,the proposed method has the good resistance to the boundary noise;on the other hand,it guarantees the accuracy of image contour without apparently impact on the contour extraction speed.