Intertwined of JPEG image

Hello, I have 2 JPEG images intertwined with each other. So, we have 4 fragments, fragment1_1, fragment1_2, fragment2_1 and fragment2_2. So, the order is fragment1_1 followed by fragment2_1, then fragment1_2 and last is fragment2_2. To distinguish fragment1_1 and fragment2_1, we can use SOI marker to identify the beginning of fragment2_1.

But, we have problem to distinguish fragment2_1 and fragment1_2 because there are no markers that can be used to recognize the consequent fragment. These is the data area (most probably compressed). I need to find a way to solve this. The idea at this moment is to decode JPEG, so we can use RGB value to calculate the distance of each line. Is it possible to decode fragmented JPEG files in matlab or can suggest other way to identify the fragmentation boundary.

Thank you

Answer 1

[Bella Sam]

Thank you for your reply. Actually what I am trying to do is to carve JPEG images from a dataset (like CSI). This dataset is an imaged from a corrupted hard disk, which all files in the hard disk will be mirrored as one single file in hex values. These images, mixed with different files and their own,can be viewed using hex editor.

There are many types of file such as word, pdf, bmp, png and so on, all in hex values. My interest is only on JPEG files. Some JPEG can carve directly because it is complete and not fragmented, but some JPEG files are fragmented, either with other types or its own. To be noted, the dataset is missing of its FAT, so we cannot allocate the files using cluster address. So, I need to recognize where the fragmentation point as I explain before. Thank you

[Image Analyst]

I don't know how to help you. If the file is fragmented, I don't know if there is something in the fragment that tells you where the prior fragment was located. Otherwise, if you don't have the directory telling you which sector each fragment lives at, and you can't daisy chain your way around, all fragments would look like gibberish. If there is some sort of pattern at the start of each fragment saying where the prior and next fragments live, then maybe you can use that. But I'm no expert in file reconstruction. It seems that if you had the middle chunk of a large jpeg image taking up a sector on the drive, and there were no clues as to what it is or what fragments it's linked to, then it's just a bunch of arbitrary incomprehensible binary junk.

By the way, Answers to questions asked in Answers should be comments to that Answer, not as new, additional Answers to your original question.

[Bella Sam]

Oh, sorry about the new comment I made. However, thanks a lot for your time and I appreciate your comment on this matter.

Thank you :)

[Image Analyst]

So are you questions all answered? If so, please mark as "Accepted".

Answer 2

[Image Analyst]

I've never heard of such a thing. I've heard of "interlaced", for example when you save a PNG image, or in the old "FlashPix" format. Are you sure there is such an option in jpeg, and if there is, if imread() can understand such a format? (Ashish Uthama would know - he's on the Mathworks image format team, maybe he'll answer or check the newsgroup where he seems to hang out more).

Or are you talking about two images that have been spliced together and then just saved as a jpeg image, which is a totally different concept than what I said first, because in this latter concept, the image is an ordinary, run of the mill jpeg format, it's just that the image has a special appearance.