1、-. - .:_. -, . . .-. / -.:_-:_-_: _ -1 -: . . . l - - . -.-1 - , -: _ ., :.:. :. :o:.;_ :-: - -_.-.:-. Alii Auoclatlon for Information and Image Management 1100 Wayne Avenue, Suite 1100 Silver Spring, Maryland 20910 Telephone 301/587-8202 ANSI/ AIIM MS44-1988 *;:=z:J Approved As December 22, 1_988 E
2、tt-:e:-:-: 1 ANSI/ AIIM MS44 1988 Standard for Information and Image Management -Recommended Practice for Quality Control of Image Scanners Association for Information and Image Management This practice provides procedures for the ongoing control of quality within a digital document image management
3、 system. The objective of the document is to provide a means of quality control frorri. input to output, i . Foreword (This foreword is not part of American National Standard for Information and Image Management-Recommended Practice for Quality Control of Image Scanners, AIIM MS44-1988.) This docume
4、nt provides procedures for the ongoing verification of the quality of a digital document image $torage and reproduction system. With the coming of electronic digital imaging systems, the need for continuing the quality concepts already established in the micrographics arena has evolved. While . a nu
5、mber of test targets are available in use for the micrographics and facsimile areas, these targets are specific to those areas and do not address the needs of the digital image management system user community. This recom mended practice addresses those needs it is proposed that the user construct h
6、is or her own target using the guidelines provided and/or obtain the scanner targets discussed in this document from AIIM. Quality control procedures such as the ones presented in this recommended practice are necessary for every user of digital image management systems. These procedures are designe
7、d to ensure that all documents entered into the . scanning system will be available for output with the necessary quality. It is recommended that these procedures be used by quality control personnel on a regular basis. Companion documents are in the process of being developed for scanning engineeri
8、ng drawings larger than USA “A!. size and for aperture card scanning. Suggestions for improvement to this practice are welcome. They should be addressed to the Chairman, AIIM Standards Board, Association for Information anci Image Management, 1100 Wayne Avenue, Suite 1100, Silver Spring, Maryland 20
9、910. At the time this standard was approved, the Standards Board of the Association for . Information and Image Management had the following members: Marilyn Courtot, Chair Thomas E. Berney Joseph Comiskey Delmar Johnson. Don Klosterboer E. Brien Lewis Alan S. Linden Wesley J. Micket Dale 0. Rupp Ge
10、orge Thoma Herbert J. White, Jr. ii The AIIM Electroruc Imaging Committee, Cl3, pro cessed and approved this recommended practice. The following were members of the comtriittee at the time of approval: Organization Information International, Inc. Access Corp. Access Corp. Applied Image AT8tf Bell La
11、bs Bell (2) To give the operator a knowledge of what the scanner can do and, more important, what it will not do; and (3) To provide the user with information needed to set up criteria for quality control procedures. The targets are sufficiently detailed so that all scan ners will fail at some point
12、. The point of failure tells the user what kinds of things the scanner will not scan prop erly (for example, colors, type size, etc.). Once the absolute capabilities of a scanner are deter mined, the. user needs to know the performance of the scanner today as compared to the last time the scanner wa
13、s calibrated. This practice points out the need for regular assessment of a given scanner compared to its own best performance. This document mainly addresses scanners for 8 1/2 inch by 11 inch (USA “A“ size) documents, however, the techniques may be useful for other scanner sizes as well. In addres
14、sing scanners for digital image management systems, this document does not address the requirt:ments of OCR scanners, color separation scanners, scanners for engineering documents larger than USA “A“ size, or other special-purpose scanners. 2. References 2.1 Referenced American National Standards AN
15、SI C16.371971, IEEE Std 167-1966 (Rl971), American National Standard Test Procedure for Facsimile. 2.2 Other Referenced Standards IEEE Std 167 A-1987, Institute of Electrical and Electronics Engineers Standard Facsimile Test Chart. The Institute of Electrical and Electronics Engineers, 345 East 47th
16、 Street, New York, NY 10017. 2.3 Referenced Publications AIIM TR2-1980, Technical Report for Information and Image Management-Glossary of Micrographics. Associa tion for Information and Image Management, 1100 Wayne Avenue, Suite 1100, Silver Spring, Md. 20910. 2.4 Related American National Standards
17、 ANSI X3.62-1987, American National Standard for In . formation Systems-Paper Used In Optical Character Recognition (OCR) Systems. . ANSI X3.86-1980 (R1987, American National Standard For Optical Character Recognition (OCR) Inks. ANSI X3.99-1983, American National Standard For In formation Systems-O
18、pticat Character Recognition (OCR) -Guidelines For OCR Print Quality. ANSI X9.7-1988, American National Standard For Bank Check Background And Convenience Amount Field. 2.5 Related PubUcations Bagg, Thomas C., Digitizing Documents: Guidelines for Image Quality, INFORM, 1(11): 6, November 1987. CCI1T
19、 Recommendation T-20 Standardized Test Chart for Facsimile Transmission, VII. 3, p.59. McCamy, C.S . On the Information in a Microphoto graph, Applied Optics, 4(4): 405, April 1965. 3. Definitions Definitions for most terms n be found in AIIM TR2-1980. The terms defined below are not found in that d
20、ocument and may be unfamiliar to the reader. Aliasing. A group of image defects, generally caused by eleme.nts of a scanned image being smaller than and/or not registered with the picture element created by the scan ner. Aliasing includes “stair-stepping“ or “jaggies“ which describe raggedness in cu
21、rves or diagonal lines. Halftone. The process of printing a continuous-tone pic ture using small black dots of varying sizes; or a picture produced by this method. Image Scanner. For the purpose of this document, a scan ner or image scanner is a device that electronically cap tures data from a docum
22、ent in a raster pattern, and that creates a digital file of the monochrome image of the document. Linearity. A measure of actual distance versus computed distance in both the x and y axis. Moire. An image defect caused by interference of one pat tern with another, which shows as a beat frequency be
23、tween the two patterns. Point. Also printers point. Measure of type size in graphic arts type. Nominally 1172 inches, precisely 0.013837 inch es (0.3514598 mm). Post-test. A verification test performed after the original documents are scanned. Pre-test. A calibration or verification test performed b
24、efore the original documents are scanned. Rectangularity. A measure of the equality of the lengths of opposite sides of a rectangle and the correctness of the angles. With perfect rectangularity, the opposite sides will be of exactly equal length and the angles will be exactly 90 degrees. Scanner. S
25、ee image scanner. Serif. In a typeface, the short decorative cross lines at the ends of the strokes in some characters. A typeface with no serifs on any characters is referred to as a sans-serif typeface. Stair-stepping. See aliasing 4. Principles of Quality Control 4.1 General Quality control is th
26、e term given to the procedures and techniques that are used to maintain consistency of output. Properly used, these procedures answer the ques- . tion, “Is what I am producing today as good as the best I can produce?“ The answer to this question is used to determine if production can continue, or if
27、 a maintenance technician must be called. A second question, “Is what I am.producing today as good as what I produced yesterday?“ may alsQ be answered, and this answer can be used to show the varia tions of a piece of equipment while still within allowable performance limits. 2 4.2 Why Do I Need Qua
28、Uty Control? In the typical digital image management system, all incoming documents are scanned, indexing information is entered, and the original paper documents are eventually destroyed. In some systems the scanned image of the docu ment may never be examined until it is needed. Strict quali ty co
29、ntrol is required to assure that the images stored are of acceptable quality and are locatable by way of the index. If a scanner is not operating properly, a large number of useless images may be stored on the system. When the problem is discovered and corrected, the original documents will have to
30、be scanned again. Procedures should be established so that any problems are discovered while the original documents are still available. The quality control procedures described in this docu ment allow the user to make sure that the system is per forming today as well as it was when originally adjus
31、ted by the manufacturer. Used on a regular basis, these pro cedures can assure the user that the scanner will produce digital images of sufficient quality for their intended use. 4.3 Establishing a Quality Reference It is extremely important to establish the definition of 40good output“ from a digit
32、al image management system. It is the nature of these systems that some images (black-and-white text) will be more faithfully recorded than others (halftones, color, etc.). Comparing a current target image against a known reference allows a nontechnical user to easily make judgments on system qualit
33、y. To establish a quality reference, the entire system must be working. The scanner should be adjusted to pro vide the best possible results for the type of material to be scanned. All of the targets, the three described in this practice, and all targets made by the user, should be scanned. The targ
34、ets should be output to hardcopy and carefully examined. If the quality of the targets is suffi cient for normal use, the hardcopies should be carefully preserved and the digital images should be permanently stored on an electronic media. All scanner adjustment set tings should be noted on the hardc
35、opy. If the normal use of the scanner is to scan all documents in the order in which they come in, without regard for content, then the scanner setup used for the referenced should reflect the best compromise of settings for the different types of orginals that will be encountered. Some users change
36、 the scanner settings for each docu ment according to the content of the document. Other users sort documents according to scanner parameters, and change the scanner settings before each batch of documents of the same type (for example correspondence scanned as one batch, pink invoice copies as a se
37、parate batch) or have separate scanners for different types of documents. In any of these cases, it is desirable to create a separate quality reference set for each type of document which will be treated separately. The user should be aware that any maintenance done to a scanner may affect the actua
38、l results of the user: controllable settings. For this reason, any time the scan ner is recalibrated by a technician a test run of all targets should be made. Once a match for the original quality of the targets has been achieved, the new proper settings should be.recorded. If the original quality o
39、f targets can not be achieved, then either the scanner still has a prob lem, or the calibration was not performed properly. In either case, additional corrective maintenance is probably necessary, (A good general rule is to run your test targets before signing off on maintenance. It is easier to mak
40、e the technician re-adjust the scanner iinmediately than to. call the technician back when you next do a test run before a scan batch.) 4.4 Pre- and Post-testing . . While it is desirable to make sure that every original document is scanned correctly, it is clearly not practical to do a test run bef
41、ore each document to be scanned. hen batches of similar documents are to be scanned, a JOOd compromise is to do a test run before the batch, eck that the scanner is properly set up and working, scan he batch, then do a test run after the batch to make sure e results are still acceptable. If the fina
42、l prescan test run and the postscan test run are the same, the scanner can be assumed to have been working properly during .the whole scanning session if the scanner adjustments were not touched during the scanning session. If the postscan test image is not acceptable, the document images in the sca
43、n batch should be examined to determine which, if any, must be scanned again. Where the scanner is adjusted separately for each original document, the pre- and post-testing approach outlined will not work. One approach for this situation Date Initials Settings 215 RTG“ 7 50 90 OK J is to determine,
44、by way of pre-testing, a small number of allowable settings which the scanner operator may use dur ing the entire scanning session (one setting per document type). This pre-testing will be done primarily with a user created target as described in Appendix A. Following the scanning session, the opera
45、tor should post-test with the same targets, using the settings established in the pre;.test run. Comparing the respective pre- and post-test runs will determine if the scanner has maintained its calibration, but canriot ensure that the proper settings were used for each document within the scanning
46、session. 4.5 Record Keeping A record (log) of all test runs should be maintained . A test run will consist of all targets which are relevant to . the type of original for which the scanneris being adjusted . This record provides management with the assurance that quality control procedures are being
47、 followed, but more important allows a technician to spot a growing problem before it becomes too serious. (An example of this would be if the log showed that the threshold setting for a proper image was consistently changing in one direc tion, then a light source may be going bad.) The log shown in
48、 Figure 1 should be completed for each te8t run. If multiple test runs are required to pro duce a satisfactory image each test run should be record . ed to assist the technician. (Appendix. B contains a blank log sheet, suitable for photoc0pying.) The number and type of settings wiU vary from scanne
49、r to scanner 5. Frequency of Testing How often test runs should be made depends on how much scanning will take place, and the consequences of improperly scanning documents. The beSt security is provided by doing a test run before and after each batch of documents scanned; where a batch is a number of documents scanned with the same :Qesults Comments INVOICES 216 JEN 12 50 95 TOO DARK -.NO Blue Statements 2/6 JEN 8 49 92 OK Blue Statements Figure 1. Test Run Log. 3 settings. This may be ten documents or ten thou
