New Steganography Method for SVG Images Based on Improved LSB Algorithm (2019) [pdf] | Hacker News Reader

New Steganography Method for SVG Images Based on Improved LSB Algorithm (2019) [pdf]

paper.ijcsns.org

This is, frankly, a terribly written paper. It’s not even described how the proposed method works? IJCSNS is also on Beall’s list of predatory publishers, which likely means that the paper was published with much weaker peer review and a high publishing fee (i.e. pay-to-publish).

I think this paper would be better titled “A Labored Effort At A Case Study Of Small-Scale Waterfall Development.”

The actual method is never described in any detail; instead the paper is chiefly concerned with the Software Development Life Cycle, requirements gathering and stakeholder communication (apparently they were surveyed using “six hat thinking” with the help of some Borland software, though who the stakeholders were in this is never explained), and so on. Figured include a class diagram (featuring 4 classes, in impossibly small print: “Steganography,” “Encryption System,” “Hiding Contents”, and... “User Registration”???) and a Gantt chart (with no labels, though we can see the whole project took about a month and a half and each phase depended on the previous one).

It's an... interesting paper, but I wish it was served over https :-)

Most of the tools I've seen were CLI only or poorly maintained, so definitely useful work.

Did the author follow through on their plan to make a GUI? This seems more like a project description than a full on whitepaper.

This paper reads like it was written by an AI trained on plagiarized undergrad assignments.

It really is odd. It keeps on defining the same acronyms (eg "Scalable Vector Graphics (SVG)") over and over again, but never defines LSB.

Wish there is more detail here. SVG is a text file format so it is hard to imagine how this would work.

Figure 1 implies that the “LSB” in question is Least Significant Bit, so my best guess at how you’d build something matching this title is to take the sort of SVG you get out of the typical GUI export (with lots of decimal places in all the paths) and twiddle the coordinates by insignificant amounts. You can squeeze a lot of bytes out of an SVG by rounding, it stands to reason that you can squeeze some bytes in by doing the opposite.

I agree. I did some more research after I wrote my comment you could indeed hide something in the floats. Prob do not need any fancy research though. Just tack on 4 numbers to each float in a path and you get some hidden space.

So you are correct!

Reply on news.ycombinator.com