Image Quality Analysis Capabilities

The image quality characteristics that Image Compliance evaluates depend on the presence of grayscale or black and white image data. When both grayscale and black and white are present, Image Compliance evaluates both, checking for all these characteristics. The evaluation of these characteristics is similar to the process which is successfully used today in the IBM® 3897 Model 4 to evaluate the quality of check images it produces.

Table 1 describes image characteristics for which Image Quality Analysis performs analysis. Image Compliance has complex algorithms to check each image for these image characteristics. The table includes reference to the Type 54 fields and names specified in ANSI Standard X9.37 Specifications for Electronic Exchange of Check and Image Data. This standard establishes the file sequences, record types, and field formats to be used for the electronic exchange of check MICR codeline data, associated check processing data, and check images in the form of cash letters. The analysis performed by Image Quality Analysis includes several of the fields specified in the Image View Analysis Record (Type 54) included in this X9.37 standard and ANS X9.100–180 Type 54 record.

Table 1. Image Compliance Capabilities
Gray Scale	Black and White	Image Compliance Capability	Type 54 Field	Field Name
Yes	Yes	Missing or torn corners	5	Partial Image
Yes	Yes	Document length	5 and 7	Partial Image or Piggyback Image
Yes	Yes	Document height	5 and 7	Partial Image or Piggyback Image
Yes	Yes	Front of document skew	6	Excessive Image Skew
Yes	Yes	Decompress failure
Yes	Yes	Image brightness	8 (X9.37 only)	Too Light or Too Dark
Yes	Yes	Image too light	8 (X9.100–180)	Too Light
Yes	Yes	Image too dark	9 (X9.100–180)	Too Dark
Yes	N/A	Image contrast	8 (X9.37 only)	Too Light or Too Dark
Yes	N/A	Image contrast	8 (X9.100–180)	Too Light
Yes	N/A	Image contrast	9 (X9.100–180)	Too Dark
Yes	N/A	Image focus		Exceeds Standard
Yes	Yes	Streaks	9 (X9.37) 10 (X9.100–180)	Streaks and Bands
Yes	Yes	Compressed data size	10 and 11 (X9.37) 11 and 12 (X9.100–180)	Below or Exceeds Image Size
N/A	Yes	Noisy image		Exceeds Standard
Yes	Yes	E13B character integrity on front	33 (X9.37) 34 (X9.100–180)	MICR Line Usability
Yes	Yes	Document dimension mismatch	7	Piggyback Image

Missing or torn corners

Analysis is performed to determine if any of the four corners of the document are folded or missing. Depending on the particular document layout, a corner that is either folded or torn may cause vital information to be missing from the image. For example, if the lower left corner of the document is folded forward over the front of the document, the routing and transit field may be partially obscured in the image. If a corner is missing from the document image and the area missing is larger than a square 0.75 inch on a side, the image may be suspect.

Document length

The length of the document may be above or below expected values. Ideally, the length, as calculated by dividing the horizontal pixel count by the pixel density (dots per inch), is within standard check length specifications. If the length of the document image is not within 4 to 9 inches, the image may be suspect. Excess length could be due to two or more documents piggy backing one another while being feed horizontally during the scanning process.

Document height

The height of the document may be above or below expected values. Ideally, the height, as calculated by dividing the vertical pixel count by the pixel density (number of dots in an inch), is within standard check height specifications. If the height of the document image is not within 2 to 4 inches, the image may be suspect. Excess height could be due to two or more documents piggy backing one another while being fed vertically during the scanning process.

Front of document skew

The document skew, defined as the measure of the angle formed between the horizontal edge of the physical document being scanned and the horizontal edge of the front of the document image, may be too great and the image may be suspect. The ideal case is when the horizontal edge of the physical document scanned is parallel to the horizontal edge of the image, meaning the skew angle is zero.

Decompress failure

An image may fail to decompress for analysis. The failure may be due to an unsupported image wrapper format or an unsupported compression algorithm. It could also be caused by corrupted compressed image data.

Image brightness, contrast, and focus

The range of picture element (pel) values and the rate of variation of picture element values within the image is analyzed by special IBM algorithms to determine the quality of the following parameters:

Image brightness

For grayscale images, the range of image gray level values may indicate the image is too dark. The brightest region of the image is not bright enough.

For black and white images, the black pixel count may indicate the image is too dark or too light.
Image contrast

The range of image gray level values may indicate the image lacks adequate contrast for legibility. Contrast analysis is not applied to black and white images.
Image focus

The variation of image gray level values may indicate the image is out of focus. Focus analysis is not applied to black and white images.

Streaks

Paper dust, toner particles, and even small pieces of paper can make even a properly bright, contrasted, and focused image unpleasing when viewed. These problems often manifest themselves as streaks of dark horizontal lines, which obscure vital document information. When localized dark areas with no contrast are detected, the image may be suspect.

Compressed data size

The image capturing device creating the delivered image performs compression. The image quality analysis process knows the compression algorithm selected and the size of the source image and can, therefore, predict the approximate compressed image size. If the size of the compressed image is too large or too small, the image may be suspect.

Noisy image

The black pixel distribution may indicate the image is too noisy, which results in the image being covered with pepper spots or blotchy patches. This analysis is not applied to gray scale images.

E13B character integrity on front

The front image may not contain well formed E13B printed characters in the codeline position (near the bottom of the image).

Codeline match

Packaged along with a document image is the MICR codeline data read from the document by the acquiring device. To match codelines, Image Compliance provides an analysis engine to compare the supplied codeline with one Image Quality Analysis recognizes from the document front image as part of the analysis process.

Document dimension mismatch

An analysis of the length and height dimensions between the front and back images to indicate the likelihood the front and back documents come from different sources.