Error level Analysis is the analysis of compression levels and errors in digital images. This form of analysis can only occur for images that undergo ‘lossy’ compression (different from ‘lossless’ compression).The most common images that undergo lossy compression are JPEGs (Joint Photographic Experts Group)

Compression is applied to many digital files i.e. images, audios, videos so as to reduce their size and enable easier transmission, save on disk space etc. Lossy compression is applied to audio, video and image files while lossless compression is normally applied to text files. Lossy compression is applied uniformly to a multimedia file resulting in uniform compression levels across the individual file. This involves the removal/discarding of small bits of information throughout the file homogeneously.

So how does Error Level Analysis work? The image in question is saved at a particular degree of compression and this saved & compressed image is compared to the original. When different levels of compression artefacts are found, it could be inferred that some parts did undergo compression more times than others or different kinds of compression have been applied to different parts of the file under question.

Recently edited parts of the image appear brighter while those that have been resaved multiple times appear darker. A comparison must be done ‘edges against edges’ and ‘surfaces against surfaces’. An image should have similar coloring while on ELA and dissimilar sections of an image should be treated suspiciously.

A downside to E.L analysis is when an image has been resaved multiple times, more resaves during the analysis will not alter the image thus there will be no discernible differences to allow for comparison with the original. Moreover, ELA has to be corroborated by other evidence and cannot be construed as undeniable evidence of image manipulation. allows for free Error Level analysis of photos and other images.

Sample image analysis:

i)Edited image

ii)ELA of image (i)

From the above, we can tell which areas of the image that were edited as areas that are most recent appear brighter than the ‘older’ areas of the image.