基于兴趣点特征的图像检索方法

针对全局图像特征无法刻画图像类别信息的缺陷, 提出一种基于兴趣点特征的图像特征检索方法. 首先对图像进行仿射 尺度不变特征转换, 并利用亮度的概率密度梯度提取兴趣点； 然后将兴趣点映射回原始图像, 采用颜色直方图作为图像特征； 最后采用相似性度量模型, 实现图像检索. 选择Corel图像库中的图像对算法性能进行实验分析. 实验结果表明, 该方法可有效提高图像的检索准确率和检索效率, 快速找到用户需要的图像.

Image Retrieval Method Based on Interest Points Features

For the global image feature could not describe the image category information, we proposed an image feature retrieval method based on interest points features. Firstly, we converted the image to affine scale invariant feature, and used the probability density gradient to extract the interest point. Secondly, we mapped these interest points back to the orginal image, and used the color histogram as the image feature. Finally, we used similarity measure model to realize image retrieval, and selected the image of the Corel image database to analyze the performance of the algorithm. The experimental results show that this method can effectively improve the accuracy and efficiency of image retrieval, and quickly find the user need image.