Applies to X9Vision?	Applies to X9Validator?	Applies to X9Assist?	\| \|	Applies to X9.37?	Applies to ACH?	Applies to CPA005?
No	No	Yes	\| \|	Yes	No	No

A limited number of image quality tests can be performed by X9Assist. These functions are enabled from the tool bar and must be set prior to an x9 file be loaded. This requirement is due to our current implementation which applies image quality tests during the file reading operation. The image quality tests that are available are as follows:

IQA (Image Quality Analysis) which will check if images are too light or too dark
ECI (Electronically Created Image) Detection which is used to identify suspect items that were computer created and were not physically scanned.

Image Quality Analysis

In the image capture environment, IQA can encompass more sophisticated tests than our more simplistic too light and too dark tests (for example, streaks, bands, edge detection, piggyback detection, and so forth). These more complex IQA functions are currently beyond the scope of what has been currently implemented within X9Assist. There are several reasons behind our current implementation and capabilities:

Too light and too dark are the minimal IQA tests performed on a standard basis by many x9 processors. There are published industry standards around IQA settings (for example, the FRB has published their too light / too dark settings on their website.
Too light and too dark tests can be performed with a minimal amount of system resources by computing pixel density. This makes the tests faster than the more complex tests. X9Assist has implemented a multi-threaded approach that results in a high performance solution that produces results in a minimal elapsed time.
Although the too light / too dark tests are simplistic, they are very effective in identifying poor images. This makes them very popular by most image capture systems.
Too light / too dark are useful tests which can catch a variety of image related errors. For example, they often detect camera issues, streaks, bands, and swapped front-back images.
Too light / too dark are obviously the easiest IQA tests to implement. X9Assist is not attempting to become a full fledged IQA product.

On a default basis, X9Assist will tag IQA failures as informational. You can change the severity of these errors by updating the appropriate messages xml configuration file.

On a default basis, X9Assist will apply the too light and too dark tests to the front image only. Examination of just the front will allow detection of typical camera issues as well as swapped front versus back issues. The front issue is also the most critical from an image exchange perspective. This is a clearly a time tradeoff, since examining the back images will typically double the amount of time required to perform these tests.

Errors are thrown during validation for items that do not meet the defined too light or too dark thresholds. When an item fails a too light or too dark test, the actual pixel density can be obtained in one of two manners:

Use the x9 record viewer (in the lower left of the dash board) to interrogate the density that was computed. The pixel density is calculated as the number of black pixels as a percentage of the total pixels that are present. You will have to slide the x9 record viewer to the right to be able to see the supplemental information for these errors.

Use the Item Viewer to examine the item, which can be launched either using Tools / Item Viewer or from the menu bar. You will then have to activate pixel density using the check box at the bottom of this panel.

Electronically Created Item Detection

ECI is used to identify suspects items that are believed to be computer (electronically) created. ECI detection may be an important consideration (especially during the onboarding of new customers) given the potential fraud associated with these items. Our inspection process implements imaging techniques which look for “perfect” images, which are typically achieved when the image is computer drawn but is otherwise not possibly achieved when images are physically captured. ECI images are identified using two separate algorithms:

Pixel noise per square inch, which is separately associated with the front and back images. Random pixel noise, especially around captured text and graphics, is typically introduced by the capture process but would not be drawn into an electronically created image.
Horizontal image skew, which is calculated from the front side image and represents any slant that may exist within the capture image. Computer drawn images do not have any skew and are very horizontally level from left to right.

ECI detection can be used identify “remotely created checks” which represent an electronic authorization to debit an account. In this transaction scenario, the image itself is computer drawn (as proxy for a customer check) and does not bear the signature of the the person on whose behalf the check is drawn. The RCC will instead include the account name or perhaps just a statement that the account holder has authorized the payment. The RCC must include a valid MICR line for the item which includes the bank routing and customer account number. RCCs are often created by credit card companies, utilities, and telemarketing operations. RCCs are processed through the x9.37 check clearing network and may be alternatively be converted to an ACH debit for clearing.

Activation

IQA/ECI can be activated in one of two manners:

IQA can be activated on a default basis using the tiff rules that are assigned to the current x9 configuration. For example, you can define certain x9 configurations to automatically perform IQA while others do not.
IQA and/or ECI can be selectively activated using the tool bar. Once activated, the selected function will be applied to the next file that is loaded and validated.

Default Settings

IQA settings are specified through the tiffRules XML configuration file. Note that you can define various tiff rules that can be automatically applied through your x9 configurations. These parameters allow you to activate IQA and then set their pixel density thresholds.