摘要
目的探讨草药图像的Hu不变矩对基于内容的维吾尔草药图像的检索效果。方法对维吾尔草药图像进行预处理,使用Canny算子、轮廓跟踪等方法获取草药图像的形状,提取草药图像的Hu不变矩形状特征,构建基于草药图像形状特征的草药图像检索系统,并进行图像检索效果验证。结果检索结果表明,基于草药图像的Hu不变矩特征对叶类、花类维吾尔草药图像的平均查准率达到70%,平均查全率达到80%。结论草药图像的Hu不变矩形状特征对基于内容的维吾尔草药图像的检索有较好的检索效果。
Objective To discuss the content-based Uighur herbal image retrieval efficiency on the herbal image Hu invariant moments. Methods Firstly, preprocessed the Uighur herbal image. Secondly, Canny operator, contour tracking and technologies alike were used to extract the shape of herbal medicine. Thirdly, the Hu invariant moments shape feature of herbal medicine images were acquired. Finally, the retrieval efficiency were validated based on the image retrieval system. Results The results show that based on Hu invariant moments feature of herbal image, the average precision rate and the average recall rate were 70 % and 80 %respectively in the search of leaf and flower shaped herbs. Conclusion content-based Uyghur herbal image retrieval can produce better results if based on the Hu invariant moments feature of herbal image.
出处
《新疆医科大学学报》
CAS
2014年第1期16-19,共4页
Journal of Xinjiang Medical University
基金
国家自然科学基金(81160182
30960097)
