Welcome to JJEM: A Multi-Disicplinary Journal of JNNCE, Shimoga
In the era of technology, multimedia contents are exposed to manipulations which leads for unsecured information transmission and reception. One such form of multimedia information is Digital Image. An image may be easily tampered in such a way that it gets difficult to find the forgery through human vision. This leads to the forgery creation in content of image data. Integrity and authenticity of image must be evaluated. In context to this, an approach for forgery detection using Superpixels segmentation is proposed. Spectral clustering is applied for images to obtain the segmentation of images in the form of Superpixels. Mapping of High dimensional feature space pixel point with proper weight computation is performed first and then K-means and normalized cuts is optimized using this objective functions. The obtained Superpixel segments forms the block for extracting the features using SIFT. The features of Superpixel segments are matched with one another to obtain the tampered part of the image based on similarity. The results obtained with the proposed copy move forgery detection method shows better precision and recall rate under various conditions than the existing forgery detection methods..
Superpixel Segmentation; spectral clustering; copy move forgery detection