Research On Image Blur Tampering Detection

Digital image forgery tampering is easier than before with the advent of the information age, the popularity of simple, practical, powerful image editing software. In recent years, disputes triggered by the fake image tampering occurs frequently in news, economic, political and other fields, which makes the image tamper detection has become an indispensable task.The most common retouch mode—blur operation in digital image in depth is studied in this paper. Currently, the study of blur tampering detection in digital image has some results, corresponding methods were put forwarded by researchers in terms of different aspects of edge, frequency and spatial domain and other fields, thus traces of blur tampering can be detected, but there are also different degrees of shortcomings and problems in these methods. In this paper, different algorithms were proposed to detect the blur operation in the spatial domain, edge features and frequency domain. The main contents are as follows:(1) For the color consistency in local neighborhood characteristics of digital cameras introduced, two algorithms of blur tampering detection in image were proposed. The algorithm is carried out an improvement on the existing algorithms. By analyzing the blur tampering operation effects of color consistency in the local neighborhood which introduction of digital camera imaging, extracting the improvement abnormal hue and hue variation features to quantitative analysis the color consistency of image in the tonal domain of HSI model, two methods of blur tampering detection were proposed, one utilizes the ambiguities and improved hue variations and the other extracts the improved abnormal hue features and send it to SVM for training. Experiments show that the detection rate of the original algorithm can be improved effectively by these improved detection algorithms, and position the fuzzy tampered regions more accurately.(2) For the edge feature problem after the blur tampering operation in digital image, a fuzzy tampering detection algorithm based on image enhancement was proposed. Firstly, the wavelet coefficients were obtained by the orthogonal wavelet transform to the digital image. Secondly, the low and high frequency wavelet coefficients were enhanced by homomorphic filtering and fuzzy sets theory, respectively. Thirdly, the reconstructed digital image with enhancement was obtained by using the inverse wavelet transform. Finally, the edge of the enhanced image was extracted by using the edge algorithms, after the normal edge was filtered by the mathematical morphology to corrosion. The experimental results show that the method can detect the blur tampering effectively, and can position the tampered regions accurately.(3) For the impact of the correlation between the image pixels and DCT coefficients by the fuzzy tampering operation, a blur tampering detection algorithm based on the correlation between the image pixels was proposed. Firstly, the mean gray value and the standard deviation of the image pixels were extracted as the correlation features; Secondly, the transition probability matrix along four directions are calculated to describe the change of the DCT coefficient and as another characteristic of the blur detection; Finally, these two type characteristics are trained by SVM, and RBF check is selected for their classification. Experimental results show that the proposed method has better detection results for blur tampering operation in images.Four methods were proposed in this paper which can effectively detect the blur tampering operation in digital image and also can locate the tampering regions, and can resist some post-processing operations effectively, such as the rotation of different angles, JPEG compression of different quality factors, with a certain degree of robustness.