关键词: 图像拼接检测, 离散余弦变换, 韦伯局部特征, 特征融合, 支持向量机

Abstract: Image splicing is one of the most popular ways to tamper image. This paper proposes an image splicing detection algorithm based on Weber Local Descriptors（WLD） in the domain of Discrete Cosine Transform（DCT）. It calculates the coefficient array by performing DCT to the detected image, and extracts WLD features to characterize the frequency distribution of the image. Support Vector Machine（SVM） is employed as the classifier to construct a prediction model and judge whether the detected image is spliced. It evaluates the proposed algorithm with standard image splicing database. Experimental results indicate that the proposed method outperforms classical image splicing detection algorithms.

Key words: image splicing detection, Discrete Cosine Transform（DCT）, Weber Local Descriptor（WLD）, feature fusion, Support Vector Machine（SVM）