色彩直方图混合特征提取在海量图像数据检索中的应用

Application of Color Histogram Mixed Feature Extraction in Massive Image Retrieval

随着多媒体技术的发展,图像的应用越来越广泛,如何有效利用图像信息具有越来越重要的意义。颜色特征不受图像大小和旋转角度的影响,鲁棒性较好,在图像检索中具有较强的应用。本文在对图像特征进行提取的基础上,提出了一种新的图像检索数据方法。该方法引入了色彩直方图、颜色中心矩、灰度直方图等颜色特征,使用色彩直方图对整幅图像的色彩信息进行统计;使用分块颜色中心矩对色彩的空间分布信息进行统计;对由于色彩浓度和对比度发生变化而对检索效果的影响,采用了灰度直方图进行淡化。通过对这三种特征距离进行加权和计算,进行相似度匹配,检索出目标图像。本文针对性地设计了基于色彩直方图混合特征的图像检索系统,实验证明色彩直方图混合特征检索的结果与示例图像在颜色与空间分布上基本相似,具有较强的实用性。

With the development of multimedia t^hnology, image application is getting more important. In image retrieval asp^t, one of the most valuable and commonly used visual features is color. On studying anti analyzing the relative technology of color feature extraction, an image retrieval method based on color feature extraction is proposed. An experiment system is designed, and experiments prove that results based on color histogram mixed features are similar to the original image in both the color tone and the spatial distribution. And it was effective for retrieval of the theme fixed image.