EP0447581A1

EP0447581A1 - Reading mechanism

Info

Publication number: EP0447581A1
Application number: EP90105224A
Authority: EP
Inventors: William H. Gunther, Jr.
Original assignee: GUNTHER INTERNATIONAL
Current assignee: GUNTHER INTERNATIONAL
Priority date: 1990-03-20
Filing date: 1990-03-20
Publication date: 1991-09-25

Abstract

A mechanism for reading data on a document having code data thereon comprising moving a document to a predetermined position. Means are provided for feeding the document from that predetermined position after a comparison is made. Means are also provided for sequentially presenting a series of comparison data to the mechanism. A reader is provided for sensing the code data on the document before the document is fed from its predetermined position. Comparing the code data on the document sensed by the reading means with the then first, at the time of comparison, sequential series of comparison data to determine whether the code data on the document matches with said comparison data. Interrupting the feeding of the document responsive to the comparing means to prevent a document from being fed if the code data on the document does not match with the comparison data.

Description

Cross-Reference

This application is a continuation-in-part of United States Patent Application Serial No. 75,125 filed July 20, 1987, which was a continuation of United States Patent Application Serial No. 899,645 filed August 25, 1986 which, in turn, was a continuation-in-part of United States Patent Application Serial No. 747,704 filed June 24, 1985; Serial No. 747,704 filed June 24, 1985, which, in turn, is a continuation of Serial No. 423,665 filed September 27, 1982. The application is also an improvement over United States Patent Nos. 4,343,129; 4,312,109; and 4,071,997 and the disclosures in said patents are incorporated herein by reference.

Description

The present invention is directed to an improved mechanism for reading a document and more particularly reading a document before feeding it.
As described and claimed in my previous patent applications, various promotions require an envelope and an insert or message enclosed with the envelope to be personalized. These applications describe and claim the advantage of reading documents before feeding them.
The mechanism of the present invention has improved capability of reading before feeding documents. The mechanism may include an optical code data reading recognition mechanism [either by Optical Mark Reading (OMR) or Bar Code Reading (BCR) or an Optical Character Recognition (OCR) or Pattern Recognition] where code data on the document is read, decoded and identified to coincide with the same data on a corresponding pre-existing comparison data means.
In existing printing machines, the forms are printed and delivered upside down so that the first form is face down in the delivery mechanism. Thus, if the hopper held one thousand sheets, sheet No. 1 would be on the bottom and sheet No. 1,000 would be on the top. When this stack of papers is processed, the next stack would have a sheet on the bottom with No. 1,001 and the last sheet on the top would be No. 2,000. Since existing folding or sealing machines must take sheets into this system face down, if such stacks were used as they were produced by the printing machine, the first stack would be processed with sheet No. 1,000 the first to be produced and sheet No. 1 the last to be produced. The next stack then would start with sheet No. 1,001 which would follow sheet No. 1 of the previous stack thereby making the entire system out of sequence.
Although this could be overcome by completing a total run of 1,000,000 forms before the first letter would be run through pre-cut folding and sealing machines and then run them in reverse order, starting with 1,000,000 back to 1 so that they could then still stay in order. The storage of the various stacks of paper would be inconvenient and could cause major problems.
The present invention overcomes these drawbacks and provides a mechanism which would process documents exactly as they are produced from a printer and feed them from the feeding devices so that the feeding devices can continue to be loaded. Before each sheet is fed out of the hopper or feeding device, it is read by a reading mechanism to verify that the sheet did, in fact, match the data on the comparison data. This ability is not available with the machines.
Furthermore, the present mechanism positions the reader so that the code data on the document is scanned and read while the document is still in a stack and before it is fed.
There are various applications for the integrity that reading before feeding gives to documents that have been computer generated either on impact or non-impact printers.
The various methods of electronic reading before feeding can be classified generally as follows:

1. Optical Character Reading
2. Bar Code Reading
3. Optical Mark Reading
4. Pattern Recognition Reading

By using one of the above mentioned methods of electronic reading/recognition, it is possible to randomly read identification code data on each document and have a computer analyze the data read and then make a decision whether to feed the sheet of paper or not feed it because it either contains the wrong code data compared to a comparison data base or is perhaps out of sequence. The basic advantage of this type of technology is to prevent feeding an improper document into a machine which would than require that the document be removed and the entire subsequent process being delayed.
An overall working understanding of the above methods and how they would be applied are as follows:

OPTICAL CHARACTER READING (OCR)

A group of characters would be placed in a predetermined location on the sheet of paper. In order to read before feeding it is necessary to scan the characters as a group while the paper is stationary, or move the reading head across the paper in order to read one character at a time. The preferred method would be to be capable of reading the group at one time and than decode them within the recognition logic and pass this information to the computer terminal which will then make decisions as to its completeness, its proper sequence order, and also what function the post processing machine should take with this particular sheet or group of sheets of paper. All of these decisions are made before the sheet of paper is fed into the next station of the post-processing machine. It is desirable, due to a possible misread, to make multiple reads of the same group of characters in order to get more than one identical consecutive read before taking action on the feeding and other machine functions.
With a moving head reader the same functions would take place, however, the mechanical movement of the head and the time to move the head across the characters make it less desirable, but in some instances it could be the only method of reading on a particular sheet.

BAR CODE READING (BCR)

With bar code reading the same basic technology is used as in the OCR method described above, however, the readable characters are transposed into a readable bar code logic, such as a UPC, Code 3 of 9, Code 2 of 5, etc. The advantages of bar code reading is that more information can be printed in the same or lesser space on the sheet of paper, plus the inherent ability to have more qualified reads in a shorter period of time.
However, with the variances in non-impact printing within any one given line, it is possible to have voids in the bar code in one or more locations within the bar. In order to avoid rejecting the entire character because of a small void, it is advantageous to raster scan the entire code several times over the entire printed code, and when three or more identical consecutive reads are made the information is passed to the control computer to make the next decision as to the feed or no feed function and how the sheet or group of sheets are to be processed.

OPTICAL MARK READING (OMR)

Optical mark reading before feeding could be used when just basic information about the sheet of paper is necessary. An example of this would be a pre-printed mark on the paper to ascertain whether or not it was in the feed hopper in its proper orientation. Since most sheet fed laser printers need the paper fed into it in an inverted (upside down) mode, it is not always possible for the operator loading the paper to make sure that it is in its proper orientation. With a pre-printed mark in a certain location the OMR reader can always give a signal to the control computer whether or not to feed the sheet of paper due to its proper or improper orientation.
Another use of OMR use would be if a multi-bin infeed device were to be used to feed the printer. The reason for a multi-bin feeder would be the ability to use various types of pre-printed forms randomly to produce a finished document. Again, since the documents would be in the input hoppers face down it is possible for the operator to load the hoppers with the improper pre-printed forms. With a unique form identification mark on each sheet is would be possible to verify that each bin had the proper pre-printed form in place before it was fed into the printer.

PATTERN RECOGNITION

Pattern Recognition is another technology whereby reading before feeding can be used as a control function to verify that the proper information is either present or absent and whether or not it is the proper information. This method is used when it is not desirable to have either readable characters or bar codes placed on a sheet of paper for aesthetic reasons. However it is necessary to verify that the proper sheet of paper is to be presented to either the printer or the post processing equipment. A simple example of this technology would be that it is desirable to print a group of sheets of paper, however, readable characters or bar codes would be objectionable. With the use of all points addressable non-impact printers, it is possible to produce graphics at the same time as characters are printed. In this instance the user could wish to identify each sheet with a different graphic symbol to make sure that the proper sequence was in fact printed. Therefore, the first sheet could have a picture of a "heart" the next a "club" the next a "spade" and the final sheet of the group a "diamond". The computer logic would ascertain if the graphic pattern it saw on the first sheet (the heart) matched the picture (pattern) it had in its memory as the first graphic it was to recognise. If they match then the control computer would give the command to feed the sheet. The control computer would then expect the next sheet to have a "club" printed on it, etc.
If any of the graphic patterns did not match to a predetermined sequence then the control computer would not feed that particular sheet and an error condition would occur.
A preferred embodiment of the invention has been chosen for purposes of illustration and description and is shown in the accompanying drawings forming a part of the specification, wherein:
Fig. 1 is a schematic view of a mechanism with which the present invention may be used.
Fig. 2 is a schematic diagrammatic view of one mechanism for using the present invention.
Fig. 3 is a schematic diagrammatic view of another mechanism for using the present invention.
Referring to Fig. 1 of the drawing which illustrates one manner of using this invention, a stack of inserts 5 is provided as well as a stack of envelopes 6. Each stack is provided in hoppers (not shown) face down. Each letter or insert assembly 5 may have a message and a personalized address thereon and may also have other codes thereon. Each envelope assembly 6 has a front panel 7 which has a personalized address, which corresponds to the personalized address on the letter assembly 5 it is to be combined with, as well as a return address. Each envelope 6 has a rear panel 8 attached to a front panel 7 by a fold line 10 and flap 9 attached to a fold line 11 to the front panel 7. The envelope assembly 6 is also provided with end flaps 60 extending from the end edges of the front panel 7. It will be noted that this envelope looks like an ordinary mailing envelope. Ideally, the insert material is first placed on the front panel 7, and the side end flaps 60 are then folded over the letter. Thereafter, the rear panel 8 is folded over the end flaps 60 and glued. Lastly, the seal flap 9 is folded over the rear panel 8 and adhered thereto so that the finished envelope looks identical to an ordinary business envelope.
The stacks of inserts 5 and envelopes 6 are preferably on either side of a pair of superimposed conveyor means 20-21. The stacks of letters 5 and envelopes 6 are fed from the bottom by any conventional feeding means, such as vacuum means, so that the insert 5 is deposited on the upper conveyor means 21 and the envelope is on the lower conveyor means 20. As has been explained above, the numbering of the inserts in the stacks are in inverse order so that additional inserts may be placed on top of it. The letters 5 and envelopes 6 have code indicia thereon and the machine is provided with reading means 23 and 24, beneath each stack of letters 5 and envelopes 6, respectively. These reading means are designed to read codes or other indicia on the letters 5 and envelopes 6 to insure that the proper envelope 6 is matched with the proper letter 5. The reading and matching is performed before the inserts and letters are fed from the stacks. The reading devices 23 and 24 may also read other indicia on letters 5 or envelopes 6 which directs the machine to have additional inserts deposited on the envelope.
The pre-cut envelope assemblies 6 and the respective pre-cut letter insert assemblies 5 are fed from the bottom of their stacks. The reading mechanisms 23 and 24 are mounted adjacent to the bottom of the stacks so that the identifying data on the letter insert assemblies and the envelope assemblies are read before they are fed from the stacks to the conveyor. If they do not match, the operator is alerted immediately, and the feeding of the insert and the letters is stopped for correction of the problem. However, the machine can continue to fold and glue previusly-fed and matched letter inserts and envelope assemblies.
The envelope 6 and the letter 5 move along by conveyors 20-21 until they reach the diverting mechanism 30. Preferably the movement of conveyors 20-21 is staggered so that both feed mechanisms call feed from the bottom of the stacks of letters 5 and envelopes 6 simultaneously.
If desired, additional feed mechanisms may be provided either behind conveyor means 20-21 or downstream thereof to feed additional pre-cut inserts to the conveyor. Furthermore, it is also possible to use a continuous web feed means 28 which will supply other inserts which have not been pre-cut but which are cut before they are fed to the conveyor system 20-21.
Immediately in front of the conveyor means 20-21 is the diverting assembly 30 comprising downward ramp mechanism 31 adapted to deflect the envelope assembly 6 downwardly onto conveyor means 40 and upward ramp means 32 adapted to move the insert assembly 5 upwardly for insertion in a folding assembly (not shown).
The conveyor means 40 comprises a continuously-moving conveyor 41 which may be driven by a wheel and chain assembly (not shown). The conveyor 41 has a plurality of transversely-located upwardly-extending longitudinally-spaced drive pins 44. The drive pins 44 are located on each side of the center line of the conveyor 41. When the envelope assemblies 6 are deposited onto the conveyor 41, the pins 44 strike the rear edge of each envelope assembly 6 to move it along with the conveyor 41. The pins 44 are adapted not only to strike and move the envelope assembly 6, but also to cause the letter assemblies 5 to be deposited on to the envelope assembly 6 and to move the envelope assembly 6 and the letter assembly 5 along together.
The folding mechanism which folds the letter assembly 5 before it is deposited onto the envelope assembly 6 may be any well known mechanism, which will fold the letter assembly 5 in two or more folds, for example, a mechanism similar to the one shown in U.S. Patent No. 1,879,990. This will fold the letter assembly 5 in two or more folds depending on the particular type of folding desired. The folding mechanism is a standard folding mechanism and, hence, will not be described in greater detail.
After the letter assembly 5 is folded, it is moved to an inclined chute 53 which overlies the conveyor 41, and terminates in a horizontal toe 54. The letter assembly slides down the chute 53 until it rests on the toe 54. Rollers (not shown) may be used to push the letters downwardly on the slide 53, if desired. The toe 54 is positioned in close adjacency over the conveyor 41. The chute 53 has longitudinal slits 55 therein to permit the pins 44 of the conveyor 41 to move therethrough. Guide rails (not shown) may also be provided to prevent the letter assembly 5 from angling when sliding down chute 53.
With the letter assembly 5 resting on toe 54, the pins 44 which are moving the envelope assembly 6 forward, move through the slits 55 and strike the rear edge of the folded letter assembly 5 so as to move it off toe 54 and deposit it on their corresponding personalized envelope assembly 6. Since the letter assembly 5 is moving at about the same speed as the envelope assembly 6, chute 53 allows letters 5 to move down to toe 54 at least us quickly as it takes the envelopes to be positioned below toe 54 so that each personalized letter assembly 5 will be deposited on its respective personalized envelope assembly 6.
The envelope assembly 6 with letter assembly 5 thereon may then be moved by pins 44 beneath an additional insert depositing assembly (not shown) where additional inserts may be deposited, if desired.
The combination may then be moved by pins 44 beneath an adhesive applicator (not shown) which applies adhesive to the envelope asembly 6. The adhesive applicator may be any conventional type of applicator, such as a sprayer, a roller, or a spotter. However, a pre-gummed envelope may also be used, if desired. The combination is then moved to an envelope wrapping assembly (not shown) which folds the envelope.
The folded and edge sealed envelope assembly 6 is then moved onto a second conveyor (not shown) which is at an angle (preferably at a right angle) to the first conveyor 41. The second conveyor moves the envelope past a flap folding mechanism (not shown) which folds the flap 9 as each envelope moves past it. An adhesive is applied to flap 9 by a suitable applicator as the envelope assembly 6 moves past it. The envelope assembly 6 is then moved past a suitable pressure roller which seals flap 9 and is then moved off the conveyor.
In Figs. 2 and 3, which illustrate other means of using this invention, a document 100 is shown in a stack 101 held in a holder 102 (shown schematically) which is located adjacent a scanning mechanism generally referred to by the reference number 105. As will be described in greater detal hereinbelow, the scanning mechanism 105 the embodiment of Fig. 2 in a lens mechanism which projects the characters 106 onto a master means 108 and the scanning mechanism 105 in the embodiment shown in Fig. 3 is a laser scanner which scans a bar code 107. In each instance, if the data is correctly read, the sheet 100 is fed from the stack 101 by a feeding mechanism 104, which has been shown greatly simplified and schematically for convenience. However, if the data is not correct, the sheet 100 is not fed and remains in the stack until the malfunction is corrected.
Referring to Fig. 2, which shows the use of the invention for OCR, the presented characters 106 on a document 100 are scanned by a scanner 105 in the form of a lens mechanism 110 to derive an array of voltages representative of the information content of incremental elements that collectively form the presented characters 106. The scanner 105 is a well-known mechanism for reading OCR and comprises a lens mechanism 110 which projects the characters 106 as images 111 onto equipment (hardware) 108 where the comparison can be made with predetermined comparison data. The equipment 108 will compare the characters (images 111) with pre-existing comparative data to determine whether they match. If there is a match, the documents 100 will be fed from the stack 101. If there is no match, the document 100 will not be fed, the system will stop so that the malfunction can be corrected. The array of representative voltages which are detected are quantized at plural quantizing levels whereby a corresponding plurality of data field geometric configurations are developed. The data field geometric configurations are developed. The data field geometric configurations are represented by binary signals and are in approximate conformity with the geometric pattern of the presented characters 106. Each data field geometric configuration represents a normalized scanned pattern, thereby compensating for nonuniformities in the presented character. The information content of discrete areas of the plural data field geometric configurations is compared in a sequential, or other feature analysis manner, with predetermined voltage characters and the compared discrete area is characterized as one of said predetermined voltage characters in response to a favorable comparison. The presented characters are identified known types when a sequence of characterization is obtained in a pre-established order associated with that known type.
The scan 105 shown in Fig. 3 is a well known system for reading bar codes 107 and comprises a bar code laser reader 120 which sends out a laser beam 121 that scans up and down the bar code 107. At the same time the laser beam 121 also sweeps across, or rasters, the bar code 107. This function allows the reader 120 to scan for a good print. The bar code 107 is scanned (rastered) several times by the laser beam 121. The scanner 105 will accept a document 100 if the scanner beam 121 makes three consecutive correct scans. If there is less than three consecutive correct scans after a predetermined interval, the scanner 105 will reject the document 200, the machine stops so that the malfunction can be corrected. If there are three consecutive correct scans, the document 100 will be fed from the stack 101 by the feeder 104. Since today's non-impact printers do not always print a solid black bar, a method of reading the bar code several times by rastering is essential. The reader is constantly looking for bars with solid print coverage.
The major uses for this type of reading before feeding, as described herein, is when documents have the need to be verified for their printing sequence, data content, and verification that no sheets were printed twice or not printed at all. These documents can be classified as documents requiring security or quality control checking, such as:

1. INSURANCE POLICIES:

Insurance policies formerly were composed with normal type setting methods, proof read for their content and accuracy, and then produced on regular printing presses. With this method, after the printing plates were made, each subsequent policy was guaranteed to be correct as to content, as it was not possible to alter the printing plate.
The major problems with the previously accepted method were that approximately 33% of the preprinted policies were discarded each year because of either subtle or major changes in federal or state laws relative to such policies, and also that it was necessary to store multiple different types of policies to meet all regulatory districts.
With the advent of computer non-impact printers it is now possible to produce policies on a daily basis as they are needed. Also, it is not necessary to have pre-printed policies for each state, but rather produce them on a current need basis on a daily basis.
Major concerns with this type of computer printing, however, were:

1.: The policies are printed on sheet fed printers and can get out of sequence.
2.: It is possible for the computer printer to have a jam and cause a page to be not printed or perhaps duplicated on a restart command from the operator.
3.: There was a need to proof read each sheet for quality control of each policy unless this could be done electronically.
4.: It is sometimes necessary to have personalized sheets from two different computer print runs merged together during the assembly of the policies and they must be in the proper sequence order.
5.: Since most of the sheets are printed (duplexed) printed on two sides it is necessary to ascertain that both sides were printed.

As these policies can vary in size from 10 pages to 150 pages, and major insurance companies can turn out several thousand policies per day, it is clear that manually verifying each page of each policy manually would be almost an undoable task.
With normal electronic reading methods, the paper document is moved past a reading head which could accept Optical Character Recognition, Bar Code Recognition, Optical Mark Recognition or Pattern Recognition. A major problem with this technique is that the reader is given only one chance to read the data. This method of reading while moving the paper past the read head can cause misreading to occur because:

1. Area scanned was not printed well enough or had voids.
2. Area scanned was not in proper position on paper.
3. One character or mark was not interpreted properly.

Any of these problems would cause that particular sheet of paper to be passed to another machine function, usually an assembly station or a conveyor, and then the entire machine would stop. The operator would then be forced to remove that sheet of paper and decide what the problem was. It is possible that two sheets have to be removed to see if the problem was one of sequence, and if that occurred then it could be necessary to remove the next sheet from the feed hopper to see if it was also out of sequence. Since all of the high speed non-impact sheet fed printers print face down, entire stacks of paper have to be removed to examine the next sheet for sequence integrity.
It is possible that an examination of the problem could take a minute or more and result in unreasonable downtime and perhaps cause more problems due to the necessary to handle large volumes of cut sheets of paper several times.
With the reading before feeding method the following advantages are inherent in the system:

1.

The date can be scanned up to 600 times per second in various areas to make sure a good read is accomplished.

2.

The data is checked against a data base to ascertain its correctness for:

a) Group integrity
b) Sequence integrity
c) Whether it was properly duplexed (printed both sides).

3.

All of the above functions are accomplished before the sheet is fed eliminating the need to stop other functions of the machine and manually removing large quantities of paper from various parts of the production process.

As these insurance policies need multiple post printing processing functions, it is desirable to have one post processing machine which can accomplish multiple tasks wither simultaneously or with an in-line production type operation. With the ability to read before feeding the sheet of paper it is possible to give instructions to very sophisticated post processing machinery to perform tasks before the sheet of paper is actually fed into the conveyor system.
An example of this would be the ability to feed from two personalized stacks of paper and interleave sheets within one policy. To accomplish this function two feeders would face each other and delivery paper to the same assembly station. With reading before feeding it would be possible to read each sequence number simultaneously and know which feeder should feed a sheet of paper into the assembly station by sequence number. This would not be possible with a machine that requires the sheet to be fed before it is read, since only after feeding would the machine know what the sequence number was and it would not know whether to switch to the other feeder.

2. PAYROLL CHECKS:

Today large producers of payroll checks have no method of verifying that the checks were actually printed and whether or not the proper amount was applied to each employee. Due to the ability to alter computer print records without too much difficulty it is possible for someone to change dollar amounts on several checks and still meet audit trail balances without ever being apprehended.
A simple example of this would be a company that had 1,000 employees and had a weekly payroll of $500,000. The common method of audit trail be to verify that 1,000 checks were printed and that the total amount computed was $500,000. If these two numbers are verified then it is usually assumed that all is in balance. However, it is very possible to have the following circumstance occur:
Employee A is due $ 700.00
Employee B is due $ 600.00
Employee C is due $ 800.00
Total 3 checks = $2,100.00
Someone with access to the computer print files can alter the records to:
Employee A gets $ 650.00
Employee B gets $ 500.00
Employee C gets $ 950.00
Total 3 checks = $2,100.00
Almost all payroll programs would balance out and the first time a problem would be detected is when Employees A & B said that their checks were for an improper amount. A look at the payroll records would ascertain this truth and they would receive the proper amount, but it would be almost impossible to find out that Employee C was paid more than he/she was due.
With the advent of non-impact printers with MICR printing capabilities it is now possible to print the check sequence number, the employee number and the dollar amount on the check all at the same time. With the ability to read all of this information before feeding it is possible to verify either on-line or subsequently back to the main data base whether the amount being paid to a particular employee was the correct amount.
A further advantage in the processing of payroll checks with a reading before feeding logic, is that payroll checks are produced for the previous weeks work. In many large organizations an employee is not longer with the company at the time the check is being produced, or an error has been identified in the calculation of the check. With the ability to read before feeding, it is possible to have the post processing equipment instructed that when that particular check is identified in the feeding station to divert it out of the normal stream without further processing. This is not possible with a mechanism with a read during feeding as the check is already on its way to additional processing before it is identified.
Also with the financial implications of large payrolls and the security problems inherent in their processing, it is desirable not to have any misreads which occur frequently in reading while feeding mechanisms.

3. BENEFIT CHECKS & EOB'S:

Similar to payroll application the mailing of benefit checks by large insurance companies and also the mailing of dividend checks by financial institutions, the same problems apply as to audit trails to verify the validity of the check amount and also the payee.
In many application of this sort multiple checks are mailed to the same payee. With the ability to read before feeding the check it is possible to see if one or more checks are to go into one envelope. With the ability to ascertain whether the next check in the feed station was to accompany the one feed before, it is now possible to assemble a group of checks to be inserted in to one envelope.

4. INVESTMENT STATEMENTS:

Presently the method of accumulating investment statements for mailing purposes by a reading while feeding method is based on reading an optical mark to indicate a beginning of group on one sheet and an end of group on the final sheet of the group. However, it is not possible to ascertain that all of the sheets were printed and actually were taken to the feeding device in the proper sequence. With reading before feeding, it is possible to actually read the account number and each sequential page number to guarantee total accuracy.

5. BANK STATEMENTS & MATCHING CHECKS:

Checks in the United States have the account number imprinted on the bottom of the check in Magnetic Ink. These numbers can be read with either magnetic readers or optically. During the banking process all of the numbers are read magnetically for debiting, crediting, and sorting purposes by the bank, however, when they are prepared for mailing back to the maker of the check by the bank along with the maker's monthly statement, they are batched together and returned to the maker with very crude merging methods. The method used for many years and still widely used, is one where a clerk places an end of batch card and the end of each check group as they are prepared for mailing. The statements are then prepared and binary bar codes are placed on the statement pages to indicate the beginning of a group along with how many checks should be accompanied with that particular statement, and then an end of group mark on the last page of the statement. The method of operation is that the statements are placed in one hopper of the mailing machine and the checks in another hopper. The statement pages are fed into the machine and as they go by the reader the beginning of statement marks are recognized and also how many checks should accompany that particular statement. Then the remaining pages of the statement are fed into the machine and the group of statements move to the next input station where the checks are assembled. The checks are then fed into the machine and counted. If on the statement page it indicated that there should be 25 checks, and 25 checks are fed into the machine before that end of group card is recognized, it is assumed the correct checks have accompanied the corresponding statement pages. The fallacy with this method is that if two clients have the same number of checks it is possible to get checks into someone else's envelope. Although the chance of this happening is quite small, and the error will soon be noticed within two or three accounts when the numbers no longer match it is a distinct possibility. A more persistent problem is that a wrong check is inserted into a group prior to the inserting operation. That way the correct count is made but it is not possible to know that one or two of the checks do not belong to that particular group.
With the reading before feeding method, it is possible to read each check before it is fed and actually compare it with an account number that can be read from the statement sheet itself. Just marking the beginning and ending of each group is not sufficient to actually identify these security type documents.

6. BILLING STATEMENTS

Billing statements can vary from one to hundreds of pages. Most retail type billing documents are one or two pages in length, however, industrial and commercial sales can vary up to hundreds of pages in length. These complex transactions have to be monitored carefully in order to verify their completeness.
Existing methods of sending these statements is either in continuous form computer printouts, or multiple individual pages. In the continuous forms method it is relatively easy to visually make sure that the first and last pages belong to the same group. However, with the advent of sheet fed non-impact printers, large numbers of pages belonging to one customer must be mailed together. It is important to read an account number and also the sequence number of each sheet to ascertain that all sheets belonging to one group were in fact grouped together, rather than just counting the number of sheets and hoping that all belonged to the same client.
Due to the variable number of sheets sent to each client it is necessary to mail them in different sizes of envelopes. Since the reading/feeding mechanism feeds in to an accumulating conveyor, which must move either to the right for one size envelope or the left for another size envelope, it is necessary to know whether the next sheet in the reader/feeder station belongs to the group currently being accumulated in the accumulating conveyor, or whether a signal should be given to the conveyor to move and allow a new group to be assembled. This can only be accomplished by reading before feeding.

7. STOCK CERTIFICATES

Stock certificates are relegated to one of the most types of security documents transferred between parties. Since the value of one certificate can amount to substantial sums of money it is necessary to verify that the correct certificate was in fact sent to the proper party. A complete audit trail must be accomplished to establish a record of continuity in the production of the certificate, the amount the certificate represents, the owner of the certificate, etc. This can only be done by reading a substantial amount of data on the certificate and matching documents, and it could be necessary to read the data more than one time for total security validation. Reading before feeding is the only method that affords this type of accuracy and control.

8. COUPON & CHECK BOOKS

Both types of documents require multiple personalized pages that could be printed at different times on different types of paper, and perhaps on different machines.
Merging these various personalized documents into a coherent group in order to make up one book is a major problem. Many times these books vary in quantity of personalized sheets from book to book. Therefore, it is necessary to know what the account number and sequence number on each sheet is to allow the proper feeding from each of the individual feeders into the accumulating hopper.
Reading before feeding is the only automated way of accumulating these books into their proper sequence and assure that each page has been placed in the proper position within the book. An example would be in a savings account check book where there will be a personalized signature page, personalized deposit pages, a group of personalized checks, then a personalized reorder page, and subsequent personalized checks. Each of these various types of documents can be in separate feeding hoppers or in some cases they could be in only one or two hoppers. However, it must be ascertained that each sheet about to be fed into the accumulating hopper does in fact belong to the previous group of sheets. If the sheet is only read when it is being fed into the hopper and a wrong number is found or perhaps it is unable to read the number on its pass by the read head, then the entire machine will stop and an operator must manually intervene to remove one or more sheets from the accumulating hopper. This of course negates much of the security of the system and allows for documents to be removed from the machine or at the least produce them out of order and delay the production of the books.

9. AUTOMATED WAREHOUSING

In large distribution centers, packages, boxes, and containers are moved automatically from one conveyor to another throughout the building in order to arrive at particular stocking bins, or to definitive loading docks for final distribution. An example of finding the proper product in a bin location by reading before feeding would be that certain items were stored in predetermined locations throughout the warehouse. There are presently electronically controlled pickers, which move about the warehouse to selected bins and retrieve one or more items from those bins. Although the automated picker goes to the proper bin to retrieve the item, there is no certainty that the item within the bin is the correct item.
With a mechanism which can read information on the item in the bin before the picker deposits it on a conveyor or within a box, it is possible to ascertain that the proper item actually was being packaged for shipment. If there is no reading device to identify the item as the proper item, it will then be deposited within a box, or onto a conveyor where it will have to be examined at a later time to make sure the shipment is correct.
This type of control is very important in all types of warehousing operations, but totally necessary with security type products such as drugs and other controlled substances.

10. MANUAL PUBLISHING

Publishing of manuals falls into many categories. Some of them could be instruction manuals, military specifications, sales manuals, training manuals, price lists, etc. Previously, these were printed in the traditional manner of producing printing plates and then printing on conventional printing presses. However, with the advent of high speed laser printers it is now possible to dynamically produce these various documents on an as needed basis. The advantages of on demand printing is that there is no inventory of outdated manuals, changes can be made rapidly with only several key strokes on a computer terminal, sectional price lists can be easily produced.
The problem, however, with dynamic printing is the assurance that all of the pages were printed, none of the pages were duplicated, and that the latest version of the manual is the one being assembled for binding and distribution. As was stated earlier, the majority of the high speed non-impact printers print face down, and therefore, it is difficult to know whether the next sheet actually belongs to the previous sheet. Also, verification of the proper pages can only be done by reading specific information on the page and comparing it with a data base for confirmation of its accuracy.
An example of the above would be in a manual of government specifications on either a military or civilian product. During the years changes and updates are made on the product and revisions have to be substituted for the older specifications. By reading definitive information on each sheet before assembling the manual, it is possible to guarantee that the proper revision pages are being collected for distribution.

11. DIRECT MAIL

In the direct mail market, it is becoming very important to incorporate more than one personalized document within the same envelope. Since the various documents are printed at different times it is necessary to have them positioned in various feed (inserting) stations on an automated mailing machine. In order to guarantee that all of the documents to the same person go into the same envelope, it is necessary to read each document before it is assembled with the previous one as the package travels through the machine. With reading before feeding it is possible to verify that all of the documents in a plurality of station do belong to the same individual before the first piece of the package is fed. Without this type of verification, the original document could travel through half of the machine before a mismatch would occur. By that time there could be a half dozen other packages also partially assembled in the machine necessitating the removal and examination of all of the packages and trying to get the system back into synchronization.

12. MAIL SORTING

At the present time all high speed mail sorting machines used either by the United States Postal Service, or by private industry sort letters by moving the letter on a high speed feeding device past a reading head which reads one character at a time. This method has an acceptance rate of only sixty (60) percent of the mail processed. Since there are billions of pieces of mail, a forty percent reject rate is very substantial and also very costly. With a method of reading before feeding it would be possible to make multiple reading attempts of the same area while electronically changing, contrast, depth of field, and also various type font logics before an actual reject occurs.
By using this method of reading before feeding the document, perhaps another sixty to seventy percent of the rejected forty percent could be accepted as good document, savings tens of millions of dollars per year.
It will thus be seen that the present invention provides an improved mechanism having the capability of reading before feeding, either by OMR, BCR, OCR or Pattern Recognition. The improved mechanism provides means for preventing further feeding of documents from a stack or repository of documents if the data that is read does not compare with the comparison data. This facilitates the problem of having to stop the machine and correct the malfunction before any more sheets are fed.

Claims

A mechanism for reading data on a document having code data thereon comprising means for presenting said document to a predetermined position, means for feeding said document from said predetermined position, means for sequentially presenting a series of comparison data to said mechanism, reading means for sensing said code data on said document before said document is fed from said predetermined position, means for comparing the code data on said document sensed by the reading means with the then first, at the time of comparison, sequential series of comparison data to determine whether the code data on said document matches with said comparison data, means for interrupting feeding of said document responsive to said comparing means to prevent a document from being fed if the code data on the document does not match with said comparison data.
A mechanism as set forth in claim 1 wherein said reading means is an Optical Character Recognition reader.
A mechanism as set forth in claim 1 wherein said reading means is a Bar Code reader.
A mechanism as set forth in claim 1 wherein said reading means is an Optical Mark reader.
A mechanism as set forth in claim 4 wherein said reading means is a Pattern Recognition reader.
A mechanism as set forth in claim 2 wherein the code data comprise characters and wherein said characters are projected by projecting means to a position where the comparison is made.
A mechanism as set forth in claim 3 wherein said code data is a bar code and wherein said bar code is scanned by a laser scanner.
A mechanism as set forth in claim 7 wherein said scanning is repeated until a correct reading is obtained.
A mechansim as set forth in claim 8 wherein the bar code is scanned until three consecutive correct readings are obtained.
A mechanism as set forth in claim 1 wherein the document comprises sheets assembled together in order to form an insurance policy.
A mechanism as set forth in claim 1 wherein the document comprises checks.
A mechanism as set forth in claim 1 wherein the document comprises investments statement.
A mechanism as set forth in claim 1 wherein the document comprises a bank statement.
A mechanism as set forth in claim 1 wherein the document wherein the document comprises a billing statement.
A mechanism as set forth in claim 1 wherein the document comprises stock certificates.
A mechanism as set forth in claim 1 wherein the document comprises coupons.
A mechanism for reading data on a document having code data thereon comprising means for holding said document in a predetermined position, means for feeding said document from said predetermined position, means for presenting comparison data to said mechanism, reading means for sensing said code data on said document before said document is fed from said predetermined position, means for comparing the code data on said document sensed by the reading means with the comparison data to determine whether the code data on said document matches with said comparison data, means for interrupting feeding of said document responsive to said comparing means to prevent a document from being fed if the code data on the document does not match with said comparison data.
A mechanism as set forth in Claim 17 wherein said reading documents are held in a stack are are fed one-by-one from said stck by said feeding means.
A mechanism as set forth in Claim 18 wherein said documents are sheets.
A mechanism as set forth in Claim 19 wherein said sheets are fed from the bottom of said stack.