Navigation : Home : Forensics



With ClearID® you get a powerful, non-destructive, easy-to-use forensic workflow that guides you through image clarification from start to SHA-1 verification, all using Adobe Photoshop®.


Note: We also offer a package called FoveaPro for Advanced Forensic Image Enhancement

The ClearID® Image Clarification Workflow

The considerations for a forensic workflow should be:

  1. that it is non-destructive
  2. all processes should be reproducible
  3. there is a consistent approach to image analysis every time
  4. the steps taken should always be documented and easy to reproduce
  5. and it should include a strategy for image verification

The ClearID Workflow incorporates all of the above considerations in one easy-to-use interface. The tabs in the interface are to be taken in order, 1 through 7. The ClearID Workflow is designed to walk you through proper image analysis with suggested tools at each step that you may or may not need to use. For example, if you image is from a progressive source and does not contain any interlaced artifacts., you would skip step 1 and proceed to step 2. The Workflow will guide you through the proper order of analysis while maintaining the integrity of your original image.

To start, type F10 and the following dialog appears:

Tab 1Here you can see the tabs numbered 1 through 7. These are the steps to be taken for the Image Clarification Workflow.

Starting with Tab #1, we address any video artifacts in the image. Specifically problems of interlace. If the image didn't come from a video source or these scene is completely static, then skip on to Tab #2, below.

There are two choices:

  • Deinterlace - pick either the even or the odd "fields" in the image.
  • Video Field Alignment - make the two fields line up and potentially double the resulting vertical resolution.

The second choice, Video Field Alignment, is a unique function added by ClearID and allows the detail of both the even and the odd fields to be combined, optionally with SUPER-RESOLUTION. Sometimes this can be the difference between reading the numbers on a license plate or not.

The second step is to improve signal-to-noise in the image:

Tab 2The best option is to perform frame averaging to allow the noise in the successive frames to cancel out and provide an image that has more tonality that can in turn have contrast enhancement applied or be sharpened more. For analog video images or for low-compression setting DV images, this can be quite effective. However for clips that have had high levels of DV compression, most of the useful (and thus self-cancelling) noise is already gone and you will see little or no improvement. ClearID can average well over 1,000 frames.

Reducing image size actually can have the side-effect of improving the image quality. If the details that you care about are large, compared to the size of a pixel, then shrinking the image will not remove the smallest important details, but WILL improve the tonality of what remains. For very noisy images, especially when frame averaging is not an option, image size reduction is an important tool.

As with all of these steps, if it does not apply, skip on to the next one. In this case, #3.

The third step is to fix a focus problem if there is one:

Tab 3The ClearID Workflow provides two methods for this, both based upon a technique called "Deconvolution." Basically this is a mathematical way to provide a corrective lens to the image to try to arrive at what the original image must have been like. This was used extensively on the pictures from the Hubble Telescope prior to the first service mission because the telescope was nearsighted.

Adobe's Smart Sharpen and ClearID's Interactive DeBlur approach deconvolution in different ways. Interactive DeBlur can yield better results for both motion blur and focus blur, but can also have a number of artifacts that are less than perfect. Smart Sharpen can yield some useful results with its "Lens Blur" setting, but these results are rarely as "sharp" as those from Interactive DeBlur. Both are provided in the hopes that ONE of them will work.


The CSI Effect:

There are a number of problems with "refocusing" any image. First, you will be less able to get a "sharp" result as the images are more noisy. Second, compression artifacts are much like noise and they interfere with deconvolution. Third, the larger the focus correction required, the worse the result will be.

(We have acheived some ery good results by applying Interactive DeBlur to a number of frames, each with the subject in different positions; because the subject moved, but the compression artifacts did not. Essentially this gave us views of different parts of the subject even when there were bad artifacts.)

This IS a powerful technique but does not work under all circumstances. When you see tiny or deeply out-of-focus images on CSI get corrected and blown up, that is fiction, not forensic science.


The fourth step is to improve the contrast of the areas you care about:

Tab 4In the left column we've added our own version of Photoshop's Levels and Curves. The major difference with them is that they use a different color model than Photoshop that will preserve colors while making the images lighter or darker. Hence, COLOR-SAFE. And, the Color-Safe Curves function is much easier-to-use. We also use the Photoshop High Pass function (well suited to isolating detail or suppressing very large brightness ranges). Finally, the "Pick a Color Channel" choice can be useful when the detail you're looking for is mostly in one color (like skin detail would best be seen in the opposing color channel -- since skin tends to be yellow/brown, the most contrast for skin would be in the blue channel).

The fifth step is to remove noise in the image, if possible:

Tab 5There can be many kinds of noise in an image. Obviously if you reduce the noise in the image, it can then be sharpened more (see the sixth step, next), but it is important that the details in the image not be the same size as the noise or you may be greatly limited in what you can achieve.

However PATTERN NOISE, either the dots you see when running the vacuum cleaner and watching television at the same time or patterns in materials, can be removed. The ClearID function Pattern Remover (shown in the dialog) is extremely good at removing patterns and leaving the underlying data intact.

No. JPEG artifacts are not "patterns" in this sense and cannot be simply removed.

The sixth step is to sharpen the resulting image:

Tab 6The main standby for sharpening images in Photoshop is the UnSharp Mask function. ClearID provides a number of others that do not suffer from the same problems that USM does:

  • Ultimate Sharpener provides as much or more sharpening than USM, but less halo effect.
  • Edge Enhancer is similar to USM, but allows you to not sharpen the noise. And you can keep just the light or dark edges.
  • Detail Sharpener leaves the edges alone and instead enhances the details in the image. No halos at all.
  • Custom Matrix Convolution has presets for horizontal or vertical sharpening which can be especially useful for video images.
  • The Equalization functions are useful for isolating fingerprints on high contrast printed material.

Sharpening is the final image processing step, but the last workflow step is verification.

The seventh and final step is to generate a Verification Report

Tab 7This step answers these questions:

  • "What steps did I perform?"
  • "How do I recreate this exact result?"
  • "How do I know this IS the exact result?

This is all possible in the ClearID Workflow because each step is a separate layer in the Photoshop file. In addition, the name of each layer contains the function performed AND the settings used for that function. Finally each layer has a SHA-1 checksum performed during the operation and the last six digits are stored in the layer name.

When the Verification Report is performed, a text file is created with details on all of the layers and a new SHA-1 checksum is performed on each layer. The old SHA-1 is still in the layer name, and a new one will show the same final six digits. The entire point is to demonstrate the exact steps performed and that anyone doing these steps will get the same SHA-1 values.

ClearID is intended to make Forensic Image Clarification straighforward, robust, and entirely repeatable.

Our goal through this process is to make Photoshop easy-to-use. ClearID answers the questions "Where do I start" and "What do I do next?"