Jump to content

  • Log In with Google      Sign In   
  • Create Account

#Actualaregee

Posted 26 November 2013 - 01:11 PM

I am not any expert in cryptography either, but I just want to toss in an immediate thought I have from looking at the resulted encryption from the original pictures.

 

If you can see patterns and average colours or other hints in the result, then you are giving away information that can indicate weaknesses in your algorithm.

 

For instance: intuitively it seems like you can sort of do an average on colour values and get the same average as you have on the original picture.  On the picture with geometric figures, you can see lots of lines.  While it is not obvious what it is supposed to be in the encrypted versions of the pictures, it gives away information that can help in determining your algorithm to attack your encryption.

 

I can tell from the white noise picture that you are dealing with a black and white picture for instance.

 

I might be wrong about my assumptions, but it does give me a starting point as to figuring out your algorithm.  To me it sounds like The Comet's suggestion is a better one, or even generate a random seed from your password, and xor a random sequence from that seed over the picture.  To decode, just generate the same random seed and run the same random values and you will have the original data. 

 

No need to say that all these suggestions are really weak encryption methods, since it is easy to break as long as you have the algorithm.  I have a feeling the same goes for your spinwheel algorithm too.

 

A small note to what "TheComet" said, that you could obfuscate a little bit to make things harder.  While that is true, obfuscating doesn't really do much to secure your data.  It is better to have a proper encryption scheme right from the start.  Read about congruences and products of massively large prime numbers to do proper encryption or use a library if you want really solid encryption.  It is advanced, but there are examples that are easy to follow how to do this with examples using small prime numbers.  The technique is the same as in "proper" encryption, but you won't achieve really good encryption unless you use massively large prime numbers.

 

Security in today's encryption algorithms lie in the fact that it is hard to find the factors from a composite number from two massively big prime numbers.  If you can find a way to factor such numbers quickly, you will render almost all encryption methods useless.  

 

EDIT: Having a second look at your images, I see that my assumptions are not entirely right, but my point is still true:  the result should look entirely random, with no patterns or colour-similarities or other hints at all.  Just pure random data.


#2aregee

Posted 26 November 2013 - 01:05 PM

I am not any expert in cryptography either, but I just want to toss in an immediate thought I have from looking at the resulted encryption from the original pictures.

 

If you can see patterns and average colours or other hints in the result, then you are giving away information that can indicate weaknesses in your algorithm.

 

For instance: intuitively it seems like you can sort of do an average on colour values and get the same average as you have on the original picture.  On the picture with geometric figures, you can see lots of lines.  While it is not obvious what it is supposed to be in the encrypted versions of the pictures, it gives away information that can help in determining your algorithm to attack your encryption.

 

I can tell from the white noise picture that you are dealing with a black and white picture for instance.

 

I might be wrong about my assumptions, but it does give me a starting point as to figuring out your algorithm.  To me it sounds like The Comet's suggestion is a better one, or even generate a random seed from your password, and xor a random sequence from that seed over the picture.  To decode, just generate the same random seed and run the same random values and you will have the original data. 

 

No need to say that all these suggestions are really weak encryption methods, since it is easy to break as long as you have the algorithm.  I have a feeling the same goes for your spinwheel algorithm too.

 

A small note to what "TheComet" said, that you could obfuscate a little bit to make things harder.  While that is true, obfuscating doesn't really do much to secure your data.  It is better to have a proper encryption scheme right from the start.  Read about congruences and products of massively large prime numbers to do proper encryption or use a library if you want really solid encryption.  It is advanced, but there are examples that are easy to follow how to do this with examples using small prime numbers.  The technique is the same as in "proper" encryption, but you won't achieve really good encryption unless you use massively large prime numbers.

 

Security in today's encryption algorithms lie in the fact that it is hard to find the factors from a composite number from two massively big prime numbers.  If you can find a way to factor such numbers quickly, you will render almost all encryption methods useless.  


#1aregee

Posted 26 November 2013 - 12:58 PM

I am not any expert in cryptography either, but I just want to toss in an immediate thought I have from looking at the resulted encryption from the original pictures.

 

If you can see patterns and average colours or other hints in the result, then you are giving away information that can indicate weaknesses in your algorithm.

 

For instance: intuitively it seems like you can sort of do an average on colour values and get the same average as you have on the original picture.  On the picture with geometric figures, you can see lots of lines.  While it is not obvious what it is supposed to be in the encrypted versions of the pictures, it gives away information that can help in determining your algorithm to attack your encryption.

 

I can tell from the white noise picture that you are dealing with a black and white picture for instance.

 

I might be wrong about my assumptions, but it does give me a starting point as to figuring out your algorithm.  To me it sounds like The Comet's suggestion is a better one, or even generate a random seed from your password, and xor a random sequence from that seed over the picture.  To decode, just generate the same random seed and run the same random values and you will have the original data. 

 

No need to say that all these suggestions are really weak encryption methods, since it is easy to break as long as you have the algorithm.  I have a feeling the same goes for your spinwheel algorithm too.

 

A small note to what "TheComet" said, that you could obfuscate a little bit to make things harder.  While that is true, obfuscating doesn't really do much to secure your data.  It is better to have a proper encryption scheme right from the start.  Read about congruences and products of massively large prime numbers to do proper encryption or use a library if you want really solid encryption.  It is advanced, but there are examples that are easy to follow how to do this with examples using small prime numbers.  The technique is the same as in "proper" encryption, but you won't achieve really good encryption unless you use massively large prime numbers.


PARTNERS