On the robustness of constrained convolutional neural networks to JPEG post-compression for image resampling detection.
Belhassen BayarMatthew C. StammPublished in: ICASSP (2017)
Keyphrases
- lossy compression
- jpeg images
- image compression
- progressive transmission
- image data
- compression rate
- input image
- convolutional neural networks
- compression algorithm
- file size
- lossy image compression
- expert group
- compression ratio
- block coding
- image quality
- multiscale
- convolutional network
- image fidelity
- compressed images
- arithmetic coder
- image coding
- compression scheme
- jpeg ls
- transform coding
- continuous tone images
- jpeg compressed images
- compression artifacts
- decoding process
- perceptual quality
- region of interest
- embedded image
- feature extractor
- lossless compression
- watermark embedding
- digital images
- perceptual distortion
- trellis coded quantization
- bits per pixel
- motion estimation
- video sequences
- quantization table
- feature extraction