<div dir="ltr"><div class="gmail_extra"><div class="gmail_quote">On Sun, Oct 26, 2014 at 6:22 AM, <span dir="ltr"><<a href="mailto:dan@geer.org" target="_blank">dan@geer.org</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><span class="">> Here's Balasubramaniyan's PhD thesis describing the Pindr0p technology:<br>
><br>
</span>> <a href="https://smartech.gatech.edu/bitstream/handle/1853/44920/balasubramaniyan_vijay_a_201108_phd.pdf" target="_blank">https://smartech.gatech.edu/bitstream/handle/1853/44920/balasubramaniyan_vijay_a_201108_phd.pdf</a><br>
<br>
Along similar lines, small noise in image acquisition is now well<br>
enough understood and discernible to say "This camera did take that<br>
picture" as it does to say "This rifle did fire that bullet," which<br>
extends to "These two pictures/bullets came from the same camera/rifle."<br><br></blockquote><div><br></div><div>How much of that image survives the conversion to a JPG compressed image.</div><div>At a RAW level this makes sense but RAW is very uncommon and most</div><div>of the raw formats are vendor specific. EXIF data can be edited... </div><div><br></div><div>Extraction of noise from image data would be helped a lot by knowing something</div><div>about the image source. An opaque cover might be enough but would require</div><div>local physical activity. Noise in a white image and noise in a black image would</div><div>likely be very different. </div><div><br></div><div>Images of clouds (like lava lamps) or leaves on trees would prove difficult to correlate</div><div>well enough to understand noise.</div><div><br></div><div>Of interest Nikon went through a recall on their D600 camera where splatters of lubrication</div><div>were showing up on the sensor over time. Sensors (shutters) were replaced and cleaned. Also </div><div>the more expensive Nikon permit saving a reference image to help remove noise and also</div><div>calibrate white balance. The reference dust image changes with time enough that it is</div><div>not a factory installed reference. </div><div><br></div><div>For images I am not convinced. It is just too easy for a bad guy to scuff an image</div><div>in difficult to detect ways. To my knowledge there is nothing kin to the yellow</div><div>dot code of copiers and printers in camera sensors (yet). Printers good enough to</div><div>worry the counterfeit folk early on were expensive enough to be engineered but</div><div>cell phone image sensors and image software has no budget room to make</div><div>hiding easy. </div><div><br></div><div><br></div><div> </div></div><br clear="all"><div><br></div>-- <br><div dir="ltr"> T o m M i t c h e l l</div>
</div></div>