基于兴趣点方向梯度直方图的图像检索方法

Image Retrieval Based on Histogram of Oriented Gradient of Interest Points

提出了一种基于图像兴趣点方向梯度直方图的检索方法。为了提高检索准确度,首先采用直方图均衡化增强图像对比度,然后利用SURF(Speeded Up Robust Features)检测子检测图像中的兴趣点,以兴趣点为中心,对兴趣点邻域内分块方向梯度直方图进行图像特征描述,最后进行相似性度量。该算法通过直方图均衡化,提取到图像中更丰富的细节信息,尤其对于颜色单一与颜色较深的图像,而且算法中充分利用了图像中包含信息量较多的图像兴趣点。实验证明。该算法提高了图像检索的准确度,相比其他算法取得了更好的检索结果。

A novel algorithm for image retrieval based on Histogram of Oriented Gradient（HOG） of interest points is presented. In order to improve the precision of the retrieval results, the contrast of image with histogram equalization is enhanced firstly and the interest points with SURF are detected. Then image features with HOG based on the blocks of the region centered at interest points are described. Finally, similar images are outputed. This algorithm extracts more detailed information through histogram equalization and better uses the interest points which contain the most information of an image. The experimental results show that this algorithm improves the precision of the retrieval results and get more satisfied retrieval results compared with other algorithms.