基于角点检测和NMF的图像Hash算法

为了有效地实现图像Hash函数在图像认证检索中的应用,提出了结合Harris角点检测和非负矩阵分解(NMF)的图像Hash算法,首先提取图像中的角点,对角点周围图像块信息进行非负矩阵分解得到表征图像局部特征的系数矩阵,进一步量化编码产生图像Hash。实验结果表明,得到的图像Hash对视觉可接受的操作如图像缩放、高斯低通滤波和JPEG压缩具有良好的稳健性,同时能区分出对图像大幅度扰动或修改的操作。

Image Hashing Based on Harris corners and NMF

In order to apply the image hash in image authentication and retrieval effectively, a new image hashing algorithm based on harris corners and nonnegative matrix factorization (NMF) is proposed, firstly harris corners are identified, around the harris corners of image block information runs NMF to obtain the coefficient matrix that capture the local features of image. Then, these features are quantized and coded to generate the hash vector. Test results indicate that the proposed method is robust against acceptable visual image scaling, Gaussian low-pass filtering and JPEG compression, at the same time is sensitive to image excessive changes or malicious tampering.