Sign My Image
 
Home Description Image Spider News Contact Help Us Cooperation References

Description

SignMyImage is single-purpose shareware which is distributed free of charge.


Image signing
SignMyImage allows you to insert your sign to the image and allows you to check if any image is signed or not. The sign can consist of letters, capitals, numbers, dashes and can be up to 10 characters long, for example: FK488x-h34

SignMyImage adds your sign directly to the image data, by slightly modifying picture intensity at a few places. The modification is so gentle that it is invisible for a human eye. SignMyImage doesn't use EXIF(or other tags) or stream properties of image files to sign the image.

SignMyImage doesn't allow you to sign already signed images. SMI can sign the following: 3 component images, which usually means RGB, but really any 3 component image, as long as none of the components is a transparency channel. Each component must be 8 bits, no more, no less. (Through experimentation, we have found some TIFs or RAWs can be 12 or 16 bits, but this is not officially supported) SMI does not support grayscale or CMYK images.


Sign verification
SignMyImage allows you to check if any image is signed or not. SignMyImage doesn't need any additional information about the image to detect the sign.


Visible watermark
Many photographers want to watermark their images by their visible signature as well. They would like to use both kind of protection - visible&invisible - together. you can have an image with both visible and invisible watermark, but the visible watermark must be written first in the blue channel. If the visible watermark is written using any other channel, the visible watermark will interfere with the insertion and detection of the invisible watermark.


Sign properties
The sign is invisible for a human eye. Each pixel, if it is gray-scale, can have a single numeric value, or if you are dealing with a 3 component image like the kind that SMI requires, each pixel will have 3 numeric values for each component. The numeric value is a measure of intensity, with 0 being the lowest intensity. If you take a random sampling of pixels and subtract the numeric values from the original and signed images, you can see some intensity changes.

Original image Signed image
Original image
Signed image

The sign stays detectable after image scaling, jpeg compression and cropping up to some level. Because of the nature of the sign coding the sign stays detectable if you PrintScreen&Paste as new image.

Very simplified rule is that the sign stays detectable in most cases when you scale the image up to 45% of its original size, when you save the image with jpeg quality up to 65%. Detectability after cropping depends on the nature of the image. The sign durability is shown on the rest of this web page.


Sign durability

1. Original images

Original image 1 Original image 2 Original image 3
Image 1 - A cave
Image 2 - Cliffs
Image 3 - An icefall
Original image 4 Original image 5 Original image 6
Image 4 - Teotihuacan
Image 5 - Alpes
Image 6 - African goats



2. Signed images

Signed image 1 Signed image 2 Signed image 3
signed & DETECTED
signed & DETECTED
signed & DETECTED
Signed image 4 Signed image 5 Signed image 6
signed & DETECTED
signed & DETECTED
signed & DETECTED



3. Cropped images
The signature is placed to several areas somewhere into the image. SignMyImage can detect the sign only if at least one (entire) area where the sign has been placed remains in the cropped image. We marked the sign areas by color rectangles at following images. The number of rectangle varies from one to five depending on the image nature.




4. Jpeg compressed images
We took signed images save them as jpegs with different quality. After that we tried to detect the sign on them.

jpeg 95% - DETECTED jpeg 90% - DETECTED jpeg 85% - DETECTED
jpeg 80% - DETECTED jpeg 75% - DETECTED jpeg 70% - DETECTED
jpeg 65% - DETECTED jpeg 60% - DETECTED
jpeg 95% - DETECTED jpeg 90% - DETECTED jpeg 85% - DETECTED
jpeg 80% - DETECTED jpeg 75% - DETECTED jpeg 70% - DETECTED
jpeg 65% - DETECTED jpeg 60% - DETECTED
jpeg 95% - DETECTED jpeg 90% - DETECTED jpeg 85% - DETECTED
jpeg 80% - DETECTED jpeg 75% - DETECTED jpeg 70% - DETECTED
jpeg 65% - NOT DETECTED jpeg 60% - NOT DETECTED
jpeg 95% - DETECTED jpeg 90% - DETECTED jpeg 85% - DETECTED
jpeg 80% - DETECTED jpeg 75% - DETECTED jpeg 70% - DETECTED
jpeg 65% - DETECTED jpeg 60% - DETECTED
jpeg 95% - DETECTED jpeg 90% - DETECTED jpeg 85% - DETECTED
jpeg 80% - DETECTED jpeg 75% - DETECTED jpeg 70% - DETECTED
jpeg 65% - DETECTED jpeg 60% - DETECTED
jpeg 95% - DETECTED jpeg 90% - DETECTED jpeg 85% - DETECTED
jpeg 80% - DETECTED jpeg 75% - DETECTED jpeg 70% - DETECTED
jpeg 65% - DETECTED jpeg 60% - DETECTED







5. Scaled images
We took signed images and scaled them with different scaling factors from 110%(zoom up) to 42%(zoom down). Results of detection are written under images.

DETECTED DETECTED DETECTED
DETECTED DETECTED DETECTED
DETECTED DETECTED DETECTED
DETECTED DETECTED






DETECTED DETECTED DETECTED
DETECTED DETECTED DETECTED
DETECTED DETECTED DETECTED
DETECTED DETECTED






DETECTED DETECTED DETECTED
DETECTED DETECTED DETECTED
DETECTED DETECTED DETECTED
DETECTED DETECTED






DETECTED DETECTED DETECTED
DETECTED DETECTED DETECTED
DETECTED DETECTED DETECTED
DETECTED DETECTED






DETECTED DETECTED DETECTED
DETECTED DETECTED DETECTED
DETECTED DETECTED DETECTED
NOT DETECTED NOT DETECTED






DETECTED DETECTED DETECTED
DETECTED DETECTED DETECTED
DETECTED DETECTED DETECTED
NOT DETECTED NOT DETECTED









Copyright © 2010 Filip Krolupper