IEEE-WVU, Anchorage 1 The Unseen Challenge Data Sets Anderson Rocha Walter Scheirer Siome Goldenstein Terrance Boult
IEEE-WVU, Anchorage 2 The Data Sets Two data sets are provided –PNG: lossless compression –JPEG: lossy compression Prevalence of images on the Internet –Sources: Google images, Yahoo Images, and Flickr
IEEE-WVU, Anchorage 3 Message Sizes For each tool, we provide four different embedding size: –Tiny: < 5% of the channel capacity –Small: > 5% & < 15% of the channel capacity –Medium: > 15% & < 40% of the channel capacity –Large: > 40% of the channel capacity For the PNG set, the message size is explicitly stated For the JPEG set, the message size is NOT stated
IEEE-WVU, Anchorage 4 Message Content Random bit sequences Snippets of mp3 songs Plain text Other images A B C
IEEE-WVU, Anchorage 5 Categories Each set consists of clean and stego images Clean set –Modified: cropping, overlay, object-appending –Non-modified: original Stego set –4 categories for JPEG, 3 categories for PNG, one for each tool
IEEE-WVU, Anchorage 6 Categories JPEG subcategories –Stego Animals Business Maps Natural Tourist Vacation –Clean Misc
IEEE-WVU, Anchorage 7 Clean Manipulated Images Object Appending Image Cropping Overlay
IEEE-WVU, Anchorage 8 PNG Tools Camaleão ( –Simple LSB insertion/modification software –Uses cyclic permutations and block ciphering to hide messages in LSBs SecurEngine ( download_4268.html) download_4268.html –Incorporates 5 crypto algorithms: Blowfish, Gost, Vernam, Cast256, and Mars –LSB encoding
IEEE-WVU, Anchorage 9 PNG Tools Stash-It ( –Windows based stego tool –Simple LSB insertion/modification software –No encryption feature
IEEE-WVU, Anchorage 10 JPEG Tools F5 ( –Resilient to 2 statistical attack –Instead of replacing LSBs directly, F5 decreases the absolute value of the DCT coefficients –Chooses DCT coefficients randomly –Matrix embedding JPHide ( –Uses blowfish to generate a stream of pseudo- random control bits to define bit encodings –Large embeddings trivial to detect
IEEE-WVU, Anchorage 11 JPEG Tools JSteg ( –40 bit RC4 Encryption –Channel capacity determination –LSB encoding in quantized DCT coefficients Outguess ( –Preserves statistics based on frequency counts –Seed based iterator available to choose embedding locations –Change minimization calculation for each seed –Remains one of the most difficult tools to detect
IEEE-WVU, Anchorage 12 PNG Data Set - Breakdown Training TinySmallMediumLarge Camaleão400 SecurEngine Stash-It Total1,1791,1871,1851,180 Non- Modified 2,000 Append- Modified 666 Crop- modified 667 Overlay- modified 667 Total4,000 4,731 total images in the PNG stego category 4,000 total images in the PNG clean category
IEEE-WVU, Anchorage 13 PNG Data Set - Breakdown Testing TinySmallMediumLarge Camaleão250 SecurEngine Stash-It250 Total ,993 total images in the PNG stego category
IEEE-WVU, Anchorage 14 JPEG Data Set - Breakdown Training F5JPHideJStegOutguess Animals1,7322, Business3, Maps3, Natural5,2111, Tourist4,9681, Vacation2, Total22,0115,3141, ,185 total images in the JPEG stego category
IEEE-WVU, Anchorage 15 JPEG Data Set - Breakdown Training Animals-Non-modified61 Business-Non-modified31 Maps-Non-modified28 Natural-Non-modified58 Tourist-Non-modified67 Vacation-Non-modified25 Misc-Non-modified1,996 Misc-Append-modified665 Misc-Crop-modified666 Misc-Overlay-modified662 Total4,259 29,185 total images in the JPEG stego category
IEEE-WVU, Anchorage 16 JPEG Data Set - Breakdown Testing TinySmallMediumLarge F5250 JPHide Jsteg Outguess Outguess Total1,6601, ,596 total images in the JPEG stego category
IEEE-WVU, Anchorage 17 Sample Usage: stegdetect JPEG Training Set Detected, CDetected, INo Steg Clean F JPHide JSteg Outguess Outguess Detected, C: correct algorithm detected Detected, I: incorrect algorithm detected Overall false detect rate for the clean image set is 8.6%
IEEE-WVU, Anchorage 18 Sample Usage: stegdetect JPEG Testing Set Detected, CDetected, INo Steg Clean F JPHide JSteg Outguess Outguess Overall false detect rate for the clean image set is 8.0%
IEEE-WVU, Anchorage 19 Sample Usage: stegdetect Detailed results for JPHide Test Set LargeMediumSmallTiny Detected, C Detected, I Negative
IEEE-WVU, Anchorage 20 Sample Usage: stegdetect Conclusions –Significant differences between the results of training and testing Weaker performance overall for testing Designed difficulty of testing set –Stegdetect performs poorly for large embeddings (non-intuitive), as well as small and tiny embeddings (expected)
IEEE-WVU, Anchorage 21 The Unseen Challenge Data Sets Lossy (JPEG) and Lossless (PNG) imagery 3 tools for PNG set, 4 tools for JPEG set 4 distinct embedding sizes for PNG, varying sizes for JPEG Clean imagery across all sets
IEEE-WVU, Anchorage 22 The Unseen Challenge Data Sets Valid approaches for use: –Detection –Detection and recovery (size or content) –Detection and destruction –Fusion No standard data set exists for steg evaluation! This set is a step in that direction!
IEEE-WVU, Anchorage 23 Download!