Denoising using Non Local Linear Filtering and Quantization Matrix Estimation Using ANFIS Algorithm for JPEG Error Analysis to Digital Image Forensics

One of the most commonly used image format is Joint Photographic Experts Group (JPEG). The recognition of JPEG compression plays a significant part in digital forensics. In previous work, JPEG image can be compressed upto n times. However, in the compression techniques noise of the JPEG images and the error analysis in the JPEG images are not primarily concentrated. Hence, the recognition of the JPEG compression results will turn out to be complicated. With the intention of overcoming these concern and eliminate the noise from the image samples, in this study formulated a blend of non local-means filter and its method noise thresholding by means of wavelets. In order to diminish the size of the JPEG image, a Growcut based seam carving technique is employed in this study. Subsequently noises are added to image to carry out Non local Linear Filter (NLF) and its Method Noise Thresholding by means of wavelets (NLFMT) denoising framework. For the purpose of assessing the influence of image compression on the performance of JPEG, a sample Discrete Cosine Transform-Singular Value Decomposition (DCT-SVD) was computed for single and double image compression, images were quantized by means of numerous quantization matrices, quantization matrix results are assessed with the help of Adaptive Neuro Fuzzy Inference System (ANFIS). Based on ANFIS, the elevated frequency coefficients in quantization matrix are employed to make a distinction among singly and doubly compressed images. Extensive experiments and evaluations with previous techniques reveal that the proposed DCT-SVD-ANFIS scheme can discover the double JPEG compression efficiently and noise in the image samples are eliminated with the help of NLFMT methods; it outperforms the existing approaches considerably based on the parameters like Peak Signal to Noise Ratio (PSNR) and Mean Square Error (MSE). The quantization matrix results were assessed using ANFIS; it has extremely much significance in the field of digital forensics.