EP1586462B1

EP1586462B1 - Method for identifying signatures in a bookbinding machine with improved sequencing error identification

Info

Publication number: EP1586462B1
Application number: EP05102936.1A
Authority: EP
Inventors: Silvano Capelli; Alfredo Cordella
Original assignee: Meccanotecnica SpA
Current assignee: Meccanotecnica SpA
Priority date: 2004-04-14
Filing date: 2005-04-14
Publication date: 2016-06-08
Anticipated expiration: 2025-04-14
Also published as: EP1586462A3; EP1586462A2

Description

Technical field

The present invention relates to the bookbinding field, and more specifically to the identification of signatures in a bookbinding machine.

Background of the invention

Bookbinding machines for the production of books from corresponding signatures are well known in the art; a typical example is that of a sewing machine, which is used to sewn together the signatures of each book. For this purpose, the signatures must be processed in a specific sequence defining the desired book.
Some solutions are known in the art (e.g. US-A-5276628 which discloses a method according to the preamble of claim 1 for automating the process of controlling the correct order of the signatures (so as to identify any sequencing error promptly). For this purpose, it is at first necessary to recognize the signatures that are processed in the sewing machine.
The recognition of the signatures can be based on the use of specific marks or bar codes. Although very accurate, this approach cannot be exploited in most practical situations; moreover, the method requires signatures with predefined characteristics so that it is not of general applicability.
Alternatively, an increasing interest is nowadays shown for the use of techniques aimed at recognizing the images printed on the signatures (for example, through suitable cameras). However, the application of these techniques in the bookbinding field is quite difficult, since the images to be recognized are indeterminate and infinite; moreover, the harsh operative conditions of the sewing machine (such as its high speed, the small internal room available, and its vibrations) involve the use of very simple devices that cannot be based on a Personal Computer (PC) architecture.
The solutions exploiting the image-recognition techniques also require a learning phase at the beginning of each job to be executed on the sewing machine. In this phase, an operator enters different configuration information defining the job, such as the total number of signatures that form each book. A sample image of every signature of the book is then acquired and stored, so as to obtain a reference sequence for the next image-recognition process. The learning phase typically involves the processing of a relatively high number of sequences of signatures (up to 30-50) for obtaining an acceptable level of accuracy; therefore, the sewing machine must work for a relatively long period without any sequencing error check.
During operation of the sewing machine, an image of the signature being currently processed is acquired and compared with the sample image of an expected signature in the sequence. Whenever an error condition is detected, the sewing machine stops and an alarm is actuated. The operator then intervenes to verify whether the error condition has actually occurred. If so, the operator must ascertain how the error condition has disarranged the order of the signatures. It is then up to the operator to determine the preferred strategy for recovering their correct order. Once a decision has been taken, the operator must rearrange the signatures accordingly; after completing the operation, the sewing machine can be restarted.
The above-described process is very time consuming, and significantly reduces the yield of the sewing machine.
Moreover, the identification of the error condition and its correction is strongly dependent on the skill of the operator. In any case, the process is prone to errors that may be due to tiredness, inattention, or distraction of the operator (even if properly skilled).
Therefore, in many practical situations the operator is unable to interpret the actual error condition that has occurred correctly. As a result, when the operator rearranges the signatures s/he may introduce further errors that make things even worse. For example, this situation may result in the production of defective books and/or in a further stop of the sewing machine.

Summary of the invention

According to the present invention, the idea of performing multiple comparisons for identifying the signatures is suggested.
Particularly, an aspect of the invention proposes a method for identifying signatures in a bookbinding machine. At first, a representation of each signature of a predetermined sequence (to be repeatedly processed by the bookbinding machine) is provided. For each cycle of the bookbinding machine (involving the processing of a current signature), the method involves the acquisition of a digital image of the current signature. The digital image of the current signature is then compared with the representation of an expected signature in the sequence. In response to a negative result of the comparison (of the digital image of the current signature), the method continues comparing the digital image of the current signature with the representation of one or more signatures that are adjacent to the expected one in the sequence. The current signature is then identified according to the results of those comparisons (of the digital image of the current signature).
In a preferred embodiment of the invention, the adjacent signatures consist of the two signatures preceding and following the expected one.
As a further enhancement, one or more further cycles of the bookbinding machine are performed after detecting the (wrong) signature; an error condition of the bookbinding machine is then determined according to the results of the comparison of a set of processed signatures (including the wrong one) with a corresponding set of expected signatures.
Typically, the bookbinding machine is stopped and an indication of the error condition is output.
A way to further improve the solution is to determine a correction action for the error condition (which is output as well).
In a specific embodiment of the invention, the solution is applied in a sewing machine (which requires a predefined number of cycles for bringing the current signature to the end of an opening station before a fixed saddle); in this case, the number of the further cycles (performed after the detection of the wrong signature) is set to such predefined number minus one.
In any case, a suggested choice for this number of further cycles is 2.
As a further enhancement, another cycle (or more) of the bookbinding machine is performed for facilitating an access to the wrong signature (for example, bringing it to the end of the opening station).
Another improvement is that of performing an idle cycle (or more) of the bookbinding machine so as to create room for inserting any missing signature.
In a preferred embodiment of the invention, the representations of the signatures of the sequence are generated from corresponding digital images of signatures that are acquired during a learning phase of the bookbinding machine. In this phase, the digital image of each signature following the first one is compared with the representation of the first signature in the sequence; when the comparison gives a positive result, the number of the signatures in the sequence can be determined according to the number of signatures that have been processed. It should be noted that this feature is also suitable to be used (alone or combined with any other of the additional features of the invention) even in a standard bookbinding machine (i.e., without the above-described solution for identifying the signatures).
A further aspect of the invention proposes a computer program for performing the method.
A still further aspect of the invention proposes a product embodying the program.
Another aspect of the invention proposes a corresponding bookbinding machine.
The characterizing features of the present invention are set forth in the appended claims. The invention itself, however, as well as further features and the advantages thereof will be best understood by reference to the following detailed description, given purely by way of a non-restrictive indication, to be read in conjunction with the accompanying drawings.

Brief description of the drawings

Figure 1 is a pictorial representation of a sewing machine in which the solution according to an embodiment of the invention is applicable;
Figure 2 shows the functional blocks of some exemplary units of the sewing machine;
Figure 3 depicts the main software components that can be used for practicing the solution according to an embodiment of the invention;
Figures 4a-4c show a diagram describing the flow of activities relating to an implementation of the solution according to an embodiment of the invention;
Figures 5a-5n are illustrative examples of different error conditions with the corresponding suggested correction actions.

Detailed description of the preferred embodiment

With reference in particular to Figure 1, a bookbinding sewing machine 100 is shown. The machine 100 is used to sew together signatures 105, in order to produce corresponding books. Each book is formed by a predefined number of signatures 105 (such as 5-50), which are ordered in a desired sequence. Each signature 105 consists of a large printed sheet being folded one or more times, so as to define several layers (each one for a page of the book). The signature 105 is identified by a corresponding image (for example, a text with or without drawings), which is printed on its front layer.
The signatures 105 are loaded into a hopper 110, with their front layers facing forwards. As described in detail in the following, an image-recognition device 115 is used to identify the signature 105 being extracted from the hopper 110 (at its bottom). The operation of extracting the signatures 105 is performed by pliers 120, which fed them in succession to an opening station 125. Each signature 105 is opened in the middle and placed astride a fixed saddle 130. The fixed saddle 130 is provided with a longitudinal slit for a chain with pushing pegs (not shown in the figure), which chain conveys the signature 105 through a shaping station 135. A throwing wheel 140 then accelerates the signature 105, which is separated from the preceding ones and placed onto a movable saddle 145 (in an open position aligned with the fixed saddle 130). The movable saddle 145 is then closed, and the signature 105 is sewn to a previous one of the current sequence by means of continuous threads. The resulting stacks of signatures 150 are then supplied to further machines (not shown in the figure) for completing the production of the books. Operation of the sewing machine 100 is managed by a programmable logic controller (PLC) 155. An operator of the sewing machine 100 interacts with the PLC 155 through an operator panel 160 (for example, including a touch-screen).
As shown in Figure 2, the image-recognition device 115 includes a camera 205 (typically formed by a set of lenses and a CCD or CMOS sensor). The camera 205 acquires a digital image representing the currently processed signature that is being extracted from the hopper. The digital image consists of a bit map that is formed by a matrix of elements (for example, with 128 rows and 640 columns). Each element of the matrix represents a basic area (or pixel) of the current signature; typically, the pixel element consists of three values (for example, each one of 8 bits) providing its Red, Green and Blue (RGB) components.
As described in detail in the co-pending application No. EP05102902.3 filed on 13 April 2005 , each digital image is preferably based on a selected Region Of Interest (ROI) of the front layer of the corresponding signature. For this purpose, there is considered a scrap area of the signature (which is typically cut by a three-knife trimmer after the sewing so as to allow opening all the pages of the book). The remaining portion of the signature defines a printable area, were it is possible to identify the printed page by removing a margin area that is typically left blank. A reduction area is now defined considering that the content of the printed page commonly starts from the top and the left (with a minimum allowable size). The region of interest is then obtained by intersecting the printed page with the reduction area.
The digital image is supplied to a control unit 210, which is formed by several blocks connected in parallel to a communication bus 215. Particularly, an image-processing unit (IPU) 220 receives the digital image from the camera 205. The image-processing unit 220 determines various parameters of the digital image (such as a high-frequency content and an average luminosity), which are used to control operation of the camera 205 automatically (for example, the focus and the exposure). The image-processing unit 220 also modifies the digital image to improve its quality (for example, applying a white-balance algorithm, correcting exposure problems, reducing a noise introduced by the sensor, and the like).
A microprocessor (µP) 225 controls operation of the image-recognition device 115. A RAM 230 is used as a working memory by the microprocessor 225, while a non-volatile memory 235 (for example, consisting of a flash E²PROM) stores information to be preserved even when a power supply is off (particularly, a control program of the microprocessor 225). An interface 240 couples the control unit 210 with the PLC 155.
The PLC 155 has a similar architecture based on a bus 255; particularly, the PLC 155 likewise includes a microprocessor 260 with a RAM 265 (used as a working memory) and a flash E²PROM 270 (storing its control program). An interface 275 interacts with the corresponding module 240 of the image-recognition device 115. A further interface 280 couples the PLC 155 with a touch-screen driver 285 of the operator panel 160.
It is emphasized that in this structure (as described in detail in the following) the image-recognition device 115 is tightly integrated with the PLC 155; such integration allows a complete automation of the process of controlling the sewing machine.
Considering now Figure 3, the main software components that run on the above-described structure (i.e., the image-recognition device 115 and the PLC 155) are denoted as a whole with the reference 300.
Particularly, a module 303 of the PLC 155 controls operation of the sewing machine. The controller 303 traces the number (Ie) of the currently expected signature in the sequence by means of a pointer 304. At the end of every operative cycle (involving the processing of a current signature, which is recognized, extracted from the hopper, and then supplied to the opening station), the number of the expected signature Ie is incremented by a unit (starting from 1); as soon as the number of the expected signature Ie reaches a total number (Nt) of signatures in the sequence (stored in a dedicated register 305), it is reset to 1 so as to start again the count for a new book: $Ie = MOD {(Ie + 1)}_{Nt}$
In a learning phase of a generic job that is executed on the sewing machine (consisting of the cycles that are required to process a first sequence of signatures), the total number of signatures Nt is set to 0, so that the number of the expected signature Ie is always incremented. On the other hand, at the end of every idle cycle of the sewing machine that does not involve the extraction of any signature from the hopper (for example, to perform a blind stitch after completing every stack of signatures, for cutting the corresponding threads, and the like), the controller 303 sets the number of the current signature Ie to a corresponding value indicating an empty position (such as 0).
At the same time, the controller 303 accordingly maintains a file 306, which stores a representation of the actual status of the sewing machine; inter alias, the machine status lists the numbers of all the signatures that are currently processed in the sewing machine from the hopper to the movable saddle (including any empty position).
During the learning phase of the job, the controller 303 continually transmits a strobe signal (for each signature to be processed) to a drive 307 of the image-recognition device 115. In response to the strobe signal, the drive 307 acquires the digital image of the current signature, which is stored into a working file 309; at the same time, the drive 307 also supplies an increment signal to a counter 312. Whenever this signal is asserted, the counter 312 likewise increments the number of the expected signature Ie (stored in a local pointer 315), which is reset to 1 after reaching the total number of signatures Nt (stored in a local register 318); in this case as well, the total number of signatures Nt is set to 0 during the learning phase (so that the number of the expected signature Ie is always incremented).
A reducer 321 receives the digital image of the current signature from the file 309; the reducer 321 synthesizes the digital image of the current signature into a reference pattern for use as a sample during a next production phase (when it is necessary to identify every signature being processed). The reference patterns of the signatures defining the desired sequence (in the correct order) are saved into a repository 324. Preferably, as described in the above-mentioned co-pending application No. EP05102902.3 , the digital image is partitioned into multiple cells (for example, of 5x5 pixel values) and a comparison parameter is calculated for each RGB component of every cell (as an average of the corresponding components in a surrounding frame). A negative threshold and a positive threshold are now determined for each comparison parameter as the lowest difference and the highest difference, respectively, with the corresponding comparison parameters of any adjacent cell (with the negative thresholds and the positive thresholds for all the boundary cells that are set to the maximum one and to the minimum one, respectively, among the corresponding thresholds associated with the non-boundary cells).
A comparator 327 receives the digital image of the current signature (from the file 309) and a selected reference pattern (from the repository 324). Particularly, during the learning phase a corresponding module 330 instructs the comparator 327 (by means of an enabling signal EN₁) always to consider the first reference pattern of the sequence (corresponding to the first signatures of the book). The learner 330 receives a match signal from the comparator 327, which signal is asserted in response to a positive result of the comparison (indicating that the current signature corresponds to the first signature of the book).
For this purpose, as described in detail in the same co-pending application No. EP05102902.3 , a set of eligible images is generated by applying any movement within an allowable range to the digital image of the current signature (with a resolution defined by half a cell). The green comparison parameters of the cells of each eligible image are calculated and compared with the corresponding green comparison parameters of the reference pattern (accumulating the differences so obtained into a cornesponding index); the eligible image with the lowest value of the difference index is selected. A representation pattern for the selected eligible image is now completed by calculating the missing red and blue comparison parameters in a similar way. An error counter is then incremented whenever the difference between each RGB component of the representation pattern (of the selected eligible image) and the corresponding RBG component of the reference pattern falls outside the range defined by the associated negative and positive thresholds. At the end of the process, if the error counter is lower than an acceptable value the digital image of the current signature is deemed to match the reference pattern (and the corresponding signal is asserted).
The learning phase ends as soon as the match signal is asserted (indicating that the current signature corresponds to the first signature of the next book). In response thereto, the learner 330 asserts an end signal and a start signal. The end signal instructs the counter 312 to set the total number of signatures Nt (in the register 318) to the current number of the expected signature Ie (provided by the pointer 315) minus 1: $Nt = Ie - 1;$
At the same time, the number of the expected signature Ie is reset to 1 (indicating the start of the new book). The end signal (together with the total number of signatures Nt in the register 318) is also provided to a settler 333 running on the PLC 155; the settler 333 copies the total number of signatures Nt into the register 305, and resets the number of the expected signature Ie in the pointer 304 to 1 (so as to synchronize the PLC 155 with the image-recognition device 115). In this way, the total number of signatures Nt is obtained automatically (thereby avoiding the need of entering the information manually).
The actual execution of the job can now begin, with the start signal that enables a verifier 342 in place of the learner 330. During this production phase, the controller 303 continues transmitting the strobe signal to the drive 307 for each operative cycle involving the processing of a signature (so as to cause it to acquire the digital image of the current signature as above). However, the verifier 342 now instructs the comparator 327 (by means of an enabling signal EN_c) to consider the reference pattern in the repository 324 that is identified by the number of the expected signature Ie (from the pointer 315); the verifier 342 receives the same match signal from the comparator 327, which signal is asserted in response to a positive result of the comparison (indicating that the current signature corresponds to the expected one in the sequence). The comparator 327 also returns the actual number (Ip) of the current signature in the sequence to the verifier 342 (which number Ip is equal to the number of the expected signature Ie when the match signal is asserted or it is set to null otherwise). Conversely, the controller 303 directly transmits a skip signal to the verifier 342 for each idle cycle (while the strobe signal remains deasserted). In response thereto, the verifier 342 sets the number of the current signature Ip to 0 (indicating the corresponding empty position).
In any case, the number of the current signature Ip and its representation pattern (when Ip<>0) are pushed into a FIFO queue 345 with a depth of 4 entries (thereby losing its oldest values); therefore, the queue 345 always provides this information for the last 4 signatures (including any empty positions) that have been processed in the sewing machine.
Whenever the match signal is deasserted (indicating that the current signature is wrong), the verifier 342 sends an error signal to the controller 303. In response thereto, the controller 303 slows-down operation of the sewing machine, so as to allow stopping it exactly after two further cycles following the completion of the current one. Therefore, when the sewing machine stops, the queue 345 will include the relevant information for the signature preceding the wrong one, for the wrong signature, and for the two signatures following the wrong one. Particularly, the numbers of those processed signatures (denoted with Ip_-1, Ip₀, Ip₁ and Ip₂, respectively) will be available only when they had been recognized (while they are set to null otherwise). As described in detail in the following, the controller 303 then sends a request signal to the verifier 342 for the information relating to each processed signature (passing the number of the corresponding expected signature Ie, which has been extracted from the file 306). In response thereto, for each unknown processed signature the verifier 342 instructs the module 327 to compare the corresponding representation pattern (from the queue 345) with the reference pattern (from the repository 324) preceding and following the one of the associated expected signature in the sequence (according to a wrap-around order); for this purpose, the verifier 342 supplies an enabling signal EN_w to the comparator 327, which signal identifies the information to be taken into account.
As in the preceding case, the comparator 327 returns the number of the processed signature Ip to the verifier 342, which stores it into the corresponding entry of the queue 345. Therefore, when the representation pattern of the processed signature matches the reference pattern of the following signature or of the preceding signature the number Ip will be equal to Ip=MOD(Ie+1)_Nt or to Ip=MOD(Ie-1) _Nt , respectively. Conversely, if the processed signature is still not recognized its number Ip remains set to null; this situation can occur when the processed signature is not an adjacent one, it is damaged, it is displaced, and so on.
The choice of considering the two adjacent signatures only has been found to be a good compromise between the opposed requirements of high accuracy and low computational complexity; in any case, it allows identifying the wrong signature in most practical error conditions (which are typically due to the swapping of adjacent signatures, to the missing of a signature, or to the repetition of the same signature). Moreover, it should be noted that the above-mentioned operations (which may be quite time consuming) are performed when the sewing machine is not working and then no time constrain applies.
The numbers of the processed signatures Ip_-1-Ip₂ are then supplied from the queue 345 to an analyzer 348 running on the PLC 155. This information provides a comprehensive representation of the real situation of the bookbinding machine, which is very useful for identifying the error condition that has occurred. The choice of performing two further cycles (after the detection of the wrong signature) allows gathering the maximum amount of information that is possible, at the same time leaving room for inserting any missing signature; moreover, it avoids placing the wrong signature on the fixed saddle (where it can fall down or be damaged if it is not opened properly). In any case, this choice has been found to be the preferred one in most practical situations; indeed, a higher number of cycles would increase the complexity of the solution without any valuable improvement of its accuracy (since the corresponding additional error conditions that might be identified are typically very difficult, if not impossible, to recover).
As it will be apparent in the following, the analyzer 348 compares the numbers of the processed signatures Ip_-1-Ip₂ with the corresponding numbers of the expected signatures (denoted with Ie_-1-Ie₂) that are extracted from the file 306, so as to determine the error condition that has occurred in the sewing machine. For this purpose, the analyzer 348 accesses a knowledge base 351. For each error type (specified by a corresponding identifier), the knowledge base 351 stores a rule and a possible command. The rule defines the occurrence of the corresponding error condition when evaluated to true; typically, the rule consists of one or more expressions that can be combined with standard logical operators (such as AND, OR and NOT). The command forces the automatic execution of additional operations (if necessary) that would facilitate the recovery of the error condition. For example, the command may force an idle cycle (or more) for inserting any missing signatures in the sequence. On the other hand, the command may force one or more further operative cycles, so as to bring the wrong signature to an ergonomic position where it is more readily accessible by the operator; typically, this operation is aimed at having the wrong signature reach the end of the opening station (before placing it on the fixed saddle).
In response thereto, the analyzer 348 (if necessary) provides a cycle signal that instructs the controller 303 to perform the required additional operation (such as the idle cycle or the further operative cycle). In any case, the analyzer 348 passes the error identifier (set to 0 when no rule in the knowledge base 351 evaluates to true) to a wizard 354. The wizard 354 accesses a repository 357, which stores a template picture for each error type (including the unknown one); each template picture represents a suggested correction action for the corresponding error condition, with possible dynamic objects defined by conesponding variables to be resolved at run-time. The wizard 354 extracts the template picture associated with the received error identifier, and replaces each variable with the corresponding actual value; for this purpose, the wizard 354 also receives the numbers of the processed signatures Ip_-1-Ip₂ (from the queue 345). The picture so obtained is then displayed on the touch-screen of the operator panel (not shown in the figure).
An operator of the sewing machine corrects the error condition (preferably following the suggestion provided by the wizard 354); for this purpose, the operator can either perform physical actions (for example, removing, inserting, replacing or moving signatures) or logical actions (for example, entering a new number of the expected signature by means of the touch-screen of the operator panel). Once the correct operation of the sewing machine has been (allegedly) restored, the operator restarts the sewing machine by means of a corresponding command A recoverer 360 intercepts the event If necessary, the recoverer 360 forces the pointer 304 (on the PLC 155) and the pointer 315 (on the image-recognition device 115) to the new number of the expected signature Ie that was entered by the operator; in any case, the recoverer 360 sends a restart signal to the controller 303 for resuming the execution of the current job.
It is evident that the suggestions provided to the operator strongly facilitate the correction of any sequencing error that has occurred in the sewing machine; more specifically, this solution simplifies the task of recovering the correct operation of the sewing machine, thereby shortening its stop period; all of the above has a beneficial impact on the yield on the sewing machine. Particularly, the identification of the error condition that has occurred is now far less dependent on the skill of the operator; moreover, factors like her/his tiredness, inattention, or distraction do not substantially affect the result of the process. This solution avoids (or at least strongly reduces) any incorrect interpretation of the error condition. Therefore, the risk of introducing further errors during the process of rearranging the signatures is reduced. As the result, it is very unlike to have defective books pass through the sewing machine, or to have further stops thereof caused by a wrong correction action. It should also be noted that the suggestions provided to the operator generally allow recovering the correct operation of the sewing machine acting at the physical level only (thereby avoiding potentially dangerous operations at the logical level).
Moving to Figures 4a-4c, the logic flow of a process that can be implemented in the above-described sewing machine is represented with a method 400. Particularly, the method begins at the black start circle 401 and then forks into two braches that are executed alternatively; a first branch consists of blocks 402-403 and a second branch consists of blocks 404-480. In both cases, the flow of activity then returns to the start circle 401 for reiterating the same operations continually (until the completion of the current job).
The first branch is executed whenever an idle cycle is required in the sewing machine (typically, at the end of every book). In this phase, the PLC at block 402 transmits the skip signal to the image-recognition device (updating the machine status accordingly). In response thereto, the image-recognition device at block 403 sets the number of the current signature Ip to 0 and inserts it into the corresponding queue (to indicate the presence of an empty position). The method then returns to the start circle 401.
On the other hand, the second branch is executed for each operative cycle of the sewing machine. In this phase, the PLC at block 404 instead transmits the strobe signal to the image-recognition device (updating the number of the expected signature Ie and the machine status accordingly). In response thereto, the image-recognition device at block 405 acquires the digital image of the current signature and likewise updates the number of the expected signature Ie (in its pointer).
The flow of activity then branches at block 406 according to the operative phase of the sewing machine. Particularly, in the learning phase the blocks 408-418 are executed, whereas in the production phase the blocks 420-480 are executed; in both cases, the method then returns to the start circle 401.
Considering now block 408 (learning phase), the first reference pattern of the sequence is extracted from the corresponding repository. A test is then made at block 410 to determine whether the digital image of the current signature matches this reference pattern. If not (meaning that the end of the first book has not been reached yet), the reference pattern of the digital image at issue is generated at block 411. Continuing to block 412, this reference pattern is appended to the end of the sequence in the corresponding repository. The method then returns to the start circle 401. Conversely, as soon as the first signature of the next book has been identified (block 410) the image-recognition device at block 414 sets the total number of signatures Nt (in the corresponding register) and resets the number of the expected signature Ie (in the corresponding pointer) accordingly. Descending into block 416, the event is notified to the PLC. In response thereto, the PLC at block 418 copies the total number of signatures Nt into the corresponding register and likewise resets the number of the expected signature Ie (in its pointer). In this case as well, the method then returns to the start circle 401. It is emphasized that in the above-described procedure a single sequence is required for completing the learning phase (so that the sewing machine exhibits a very high operative response).
With reference instead to block 420 (production phase), the reference pattern identified by the number of the expected signature Ie is extracted from the corresponding repository. The digital image of the current signature is now compared with this reference pattern at block 422, in order to determine its number Ip (which is set to null when the current signature is not recognized to be the expected one). In any case, the relevant information relating to the current signature is added to the corresponding queue at block 424.
The flow of activity then branches at block 426 according to the result of the above-mentioned comparison. If the current signature has been correctly recognized, the flow of activity returns to the start circle 401 directly; conversely, when the current signature is wrong the blocks 430-480 are executed and the method then returns to the start circle 401 as well.
Considering now block 430 (wrong signature), the image-recognition device transmits the error signal to the PLC. In response thereto, the PLC at block 432 slows-down operation of the sewing machine. The method continues to block 434, wherein the PLC causes the execution of a further cycle by transmitting the corresponding signal to the image-recognition device; particularly, the signal can be either the skip signal when the book has been completed or the strobe signal otherwise (with the number of the expected signature Ie and the machine status that are updated accordingly, if necessary). The image-recognition device at block 436 performs a further idle cycle in response to the skip signal (by repeating the same operations described above at blocks 402-403) or a further operative cycle in response to the strobe signal (by repeating the same operations described above at blocks 420-424); as a result, the number 0 or the number of the next signature, respectively, is queued up to the one of the wrong signature. Likewise, the PLC at block 438 causes the execution of another cycle by transmitting the required skip signal or strobe signal to the image-recognition device. As in the preceding case, the image-recognition device at block 440 performs another idle cycle or operative cycle, thereby adding the number 0 or the number of the further next signature, respectively, to the corresponding queue. The PLC then stops operation of the sewing machine at block 442.
A loop is then performed for each processed signature (starting from the last one). The loop begins at block 444, wherein the PLC requests the corresponding information to the image-recognition device. In response thereto, the image-recognition device at block 446 verifies whether the processed signature is known (i.e., its number Ip in the queue is different from null). If not, the blocks 448-460 are executed, and the method then passes to block 462; conversely, the same point is reached directly from block 446. Considering now block 448 (unknown signature), the reference pattern preceding the one identified by the number of the corresponding expected signature Ie (received from the PLC) is extracted from the corresponding repository. A test is now made at block 450 to determine whether the representation pattern of the processed signature (saved in the corresponding entry of the queue) matches this reference pattern. If so, the processed signature is recognized to be the one preceding the expected signature in the sequence, and its number Ip is set accordingly at block 452; the method then descends into block 460. Conversely, the reference pattern following the one identified by the number of the expected signature Ie is extracted from the corresponding repository at block 454. A similar test is made at block 456 to determine whether the representation pattern of the processed signature matches this further reference pattern. If so, the processed signature is recognized to be the one following the expected signature in the sequence, and its number Ip is set accordingly at block 458; the method then continues to block 460. The number of the processed signature Ip so obtained (at block 452 or at block 458) is now saved into the queue at block 460. The method then passes to block 462; the same point is instead reached from block 456 directly when the processed signature has not been recognized (so that its number Ip in the queue remains null). A test is now made at block 462 to determine whether the loop has been performed four times. If not, the flow of activity returns to block 444 to repeat the same operations for the preceding processed signature.
Conversely, the loop ends and the image-recognition device at block 463 returns the numbers of the processed signatures Ip_-1-Ip₂ so obtained to the PLC. In response thereto, the PLC at block 464 evaluates the different rules stored in its knowledge base. Continuing to block 465, the type of error that has occurred in the sewing machine is determined according to the results of those evaluations. A test is now made at block 466 to verify whether the command associated with the determined error type requires the execution of any additional operation (for facilitating the recovery of the error condition). If so, the PLC forces this additional operation at block 468 (resulting in the execution of an idle loop or of a further operative cycle); the method then continues to block 470. The same point is also reached from block 466 directly otherwise.
The PLC at block 470 now causes the sewing machine to enter an alarm condition (for example, switching on a warning device and automatically opening a protection carter for allowing the operator to access the opening station). At the same time, the PLC at block 472 extracts the template picture associated with the identified error type (from the corresponding repository) and resolves any variable (according to the numbers of the processed signatures Ip_-1-Ip₂). The resulting picture showing the suggested correction action is then displayed at block 474.
As soon as the operator has corrected the error condition (either physically or logically), s/he restarts the sewing machine. The flow of activity then branches at block 476 according to whether the number of the expected signature Ie has been updated. If not, the method returns to the start circle 401 directly. Conversely, the PLC at block 478 forces the pointer storing the number of the expected signature Ie accordingly. The same operation is also performed on the image-recognition device at block 480. In this case as well, the method then goes back to block 401.
A series of use cases illustrating a practical implementation of the above-described solution is depicted in Figures 5a-5n. More specifically, each figure shows a table listing the numbers of the processed signatures Ip_-1-Ip₂ with the corresponding numbers of the expected signatures Ie_-1- Ie₂; the number of the wrong signature (Ip₀) is pointed out by means of a gray background. The picture providing the suggested correction action for recovering the error condition determined by this data is shown sideways.
For example, Figure 5a relates to an error condition caused by a repeated signature (number 2) replacing the next one (number 3) in the sequence. The error condition is identified by the assertion of the following rule: $({Ip}_{0} = {Ip}_{- 1}) AND ({Ip}_{1} = {Ie}_{1}) AND ({Ip}_{1} < > 0);$
In this case, a further operative cycle (involving the processing of the signature number 6) is performed for bringing the wrong signature to the end of the opening station. The picture showing the suggested correction action then indicates that the wrong signature (number 2) should be replaced with the correct one (number 3).
Figure 5b instead relates to an error condition caused by a repeated signature (number 2) in the sequence. The error condition is identified by the assertion of the following rule: $({Ip}_{0} = {Ip}_{- 1}) AND ({Ip}_{1} = {Ie}_{0}) AND ({Ip}_{2} < > {Ie}_{2}) AND ({Ip}_{1} < > 0);$
In this case, no additional operation is required for facilitating the recovery of the error condition. The picture showing the suggested correction action then indicates that the wrong signature (number 2) should be removed, the next signatures (numbers 3 and 4) should be moved forward, and the first signature in the hopper (number 5) should be extracted.
Figure 5c now relates to an error condition caused by a missing signature (number 3) in the sequence. The error condition is identified by the assertion of the following rule: $({Ip}_{1} = {Ie}_{2}) AND ({Ip}_{0} = {Ie}_{1}) AND ({Ip}_{1} < > {Ie}_{0}) AND ({Ip}_{2} < > {Ie}_{2}) AND ({Ip}_{1} < > 0);$
In this case, an idle cycle is performed to clear the position required for the insertion of the missing signature. The picture showing the suggested correction action then indicates that the preceding signatures ( numbers 4, 5 and 6) should be moved backward and the missing signature (number 3) should be inserted.
On the other hand, Figure 5d relates to an error condition caused by the inversion of two adjacent signatures (numbers 3 and 4) in the sequence. The error condition is identified by the assertion of the following rule: $({Ip}_{0} = {Ip}_{1}) AND ({Ip}_{1} = {Ie}_{0}) AND ({Ip}_{1} < > 0);$
In this case, a further operative cycle is performed for facilitating the recovery of the error condition. The picture showing the suggested correction action then indicates that the wrong signatures (numbers 3 and 4) should be swapped.
Likewise, Figure 5e relates to an error condition caused by the last but one signature in the sequence (number 9) that is repeated instead of the last signature (number 10). The error condition is identified by the assertion of the following rule: $({Ip}_{0} = {Ip}_{- 1}) AND ({Ip}_{2} = {Ie}_{2}) AND ({Ip}_{1} = 0);$
In this case, a further operative cycle is performed for bringing the wrong signature to the end of the opening station. The picture showing the suggested correction action then indicates that the wrong signature (number 9) should be replaced with the correct one (number 10).
Figure 5f instead relates to an error condition caused by the last but one signature in the sequence (number 9) that is repeated. The error condition is identified by the assertion of the following rule: $({Ip}_{0} = {Ip}_{- 1}) AND ({Ip}_{2} = {Ie}_{0}) AND ({Ip}_{1} = 0);$
In this case, no additional operation is required for facilitating the recovery of the error condition. The picture showing the suggested correction action then indicates that the wrong signature (number 9) should be replaced with the signature (number 10) following the empty position for the blind stitch, and the first signature in the hopper (number 1) should be extracted.
Figure 5g now relates to an error condition caused by the last signature in the sequence (number 10) that is missing. The error condition is identified by the assertion of the following rule: $({Ip}_{0} = {Ie}_{2}) AND ({Ip}_{2} < > {Ie}_{0}) AND ({Ip}_{2} < > {Ip}_{0}) AND ({Ip}_{1} = 0);$
In this case, an idle cycle is performed to clear the position for the insertion of the missing signature. The picture showing the suggested correction action then indicates that the signature after the empty position for the blind stitch (number 2) should be moved backward, this cleared position should be taken by the signature before the same empty position (number 1), and the missing signature (number 10) should be inserted at this further cleared position.
On the other hand, Figure 5h relates to an error condition caused by the inversion of the last signature in the sequence (number 10) with the first signature in the next sequence (number 1). The error condition is identified by the assertion of the following rule: $({Ip}_{0} = {Ie}_{2}) AND ({Ip}_{2} = {Ie}_{0}) AND ({Ip}_{1} = 0);$
In this case, a further operative cycle is performed for facilitating the recovery of the error condition. The picture showing the suggested correction action then indicates that the wrong signatures (numbers 10 and 1) should be swapped.
Moreover, Figure 5i relates to an error condition caused by a repeated signature (number 4) replacing the preceding one (number 3) in the sequence. The error condition is identified by the assertion of the following rule: $({Ip}_{0} = {Ip}_{- 1}) AND ({Ip}_{1} = {Ie}_{1}) AND ({Ip}_{1} < > 0);$
In this case, a further operative cycle is performed for bringing the wrong signature to the end of the opening station. The picture showing the suggested correction action then indicates that the wrong signature (number 4) should be replaced with the correct one (number 3).
Figure 51 instead relates to an error condition caused by the first signature in the sequence (number 1) that is repeated instead of the last signature in the preceding sequence (number 10). The error condition is identified by the assertion of the following rule: $({Ip}_{0} = {Ie}_{2}) AND ({Ip}_{2} < > {Ie}_{0}) AND ({Ip}_{0} = {Ip}_{2}) AND ({Ip}_{1} = 0);$
In this case, a further operative cycle is performed for bringing the wrong signature to the end of the opening station. The picture showing the suggested correction action then indicates that the wrong signature (number 1) should be replaced with the correct one (number 10).
Conversely, Figure 5m relates to an error condition caused by an unknown signature. The error condition is identified by the assertion of the following rule: $({Ip}_{0} = Null);$
In this case, no additional operation is performed. The corresponding picture that is displayed simply indicates that the signature that has caused the error condition (denoted with X) has not been recognized.
At the end, Figure 5n relates to an error condition with all the relevant signatures that are unknown. The error condition is identified by the assertion of the following rule: $({Ip}_{0} = Null) AND ({Ip}_{1} = Null) AND ({Ip}_{2} = Null);$
In this case, no additional operation is performed. The corresponding picture that is displayed simply indicates that the correct sequence cannot be reconstructed.

Modifications

Naturally, in order to satisfy local and specific requirements, a person skilled in the art may apply to the solution described above many modifications and alterations. Particularly, although the present invention has been described with a certain degree of particularity with reference to preferred embodiment(s) thereof, it should be understood that various omissions, substitutions and changes in the form and details as well as other embodiments are possible; moreover, it is expressly intended that specific elements and/or method steps described in connection with any disclosed embodiment of the invention may be incorporated in any other embodiment as a general matter of design choice.
Particularly, similar considerations apply if the sewing machine has a different architecture or is based on equivalent units. In any case, the solution of the present invention can also be implemented in a gathering machine (which is used to group the signatures in the correct order from hoppers each one loaded with signatures of the same type). In this case, whenever the digital image of the signature that is being extracted from each hopper does not match the reference pattern of the expected signature (that should be provided by the hopper), the representation pattern of the (wrong) signature is compared with the reference pattern of the expected signatures in the adjacent hoppers. As a result, it is possible to determine the actual error condition that has occurred in the gathering machine (for example, when the hopper was loaded by mistake with the signatures intended for an adjacent hopper). Similar considerations apply to a perfect-binding machine, or more generally to any other bookbinding machine involving the processing of the signatures according to predefined sequences.
Likewise, the image-recognition device can have another structure or it can include equivalent components. Moreover, although in the preceding description reference has been made to a specific image-recognition algorithm, this is not to be intended as a limitation. Indeed, the proposed solution is suitable to be implemented by representing each signature with an equivalent digital image (for example, in the luminance/chrominance space), or by taking into consideration a different region of interest of the signatures (including the whole pages printed on the top layers); moreover, it is possible to store any other representation of the signatures for their recognition (up to the entire digital images), to use other comparison criteria for verifying the match of the digital image of each signature (or of its representation pattern) with the reference pattern of the corresponding expected signature, and the like.
It should be readily apparent that the principles of the invention are not limited to the described error conditions with the corresponding suggested correction actions.
In a similar manner, the proposed pictures to be displayed for facilitating the correction of the error conditions are merely illustrative; any other indication of the error conditions and/or of the correction actions are contemplated (for example, based on textual explanations, voice messages, and so on).
Similar considerations apply if the sewing machine has another size, and thus a different number of additional (operative and/or idle) cycles must be performed for bringing the wrong signature to the desired position in the opening station.
Moreover, an implementation wherein two or more operative cycles are performed for facilitating the access to the wrong signature is contemplated.
Likewise, the execution of two or more idle cycles after the identification of the error condition is not excluded.
Similar considerations apply if the programs are executed under the control of an equivalent data processing system, if they are structured in a different way, or if additional modules or functions are provided; likewise, the different memory structures can be of different types, or can be replaced with equivalent entities (not necessarily consisting of physical storage media). Moreover, the proposed solution can implement an equivalent method (for example, with similar or additional steps).
In any case, it is possible to distribute the programs in any other computer readable medium; the medium can be any apparatus suitable to contain, store, communicate, propagate, or transport the programs (for example, of the electronic, magnetic, optical, electromagnetic, infrared, or semiconductor type).
Moreover, it will be apparent to those skilled in the art that the additional features providing further advantages are not essential for carrying out the invention, and may be omitted or replaced with different features.
For example, in an alternative embodiment of the invention the digital image of the wrong signature is compared with the reference pattern of two or more signatures preceding and/or following the expected one (thereby allowing the identification of further error conditions but at the cost of an increased complexity).
In any case, a basic implementation wherein no further additional cycle is performed after identifying the wrong signature is not excluded (even if it is far less advantageous).
In some specific situations, it is also possible to avoid stopping the sewing machine in response to the identification of the error condition (for example, when its recovery is very burdensome and/or the error condition does not impair the integrity of the next books); in this case, the resulting stack of signature can be simply marked as defective so as to discard it later on.
Without departing from the principles of the invention, a simplified version of the proposed solution might output the indication of the error condition only (without suggesting any correction action).
Alternatively, the provision of any other number of further cycles of the bookbinding machine after the detection of the wrong signature is feasible (down to a single one).
In any case, the additional operations for facilitating the correction of the error condition are not essential for putting into practice the teaching of the present invention; in other words, the sewing machine can stop without performing any further operative cycle and/or any further idle cycle.
Moreover, there is nothing to prevent the implementation of the proposed solution in a sewing machine wherein the total number of signatures is input by the operator manually at the beginning of every job.
Similar considerations apply if the programs are provided in any other form directly loadable into the working memories of the relevant devices (for example, if they are transmitted to the sewing machine through a network).
At the end, the solution according to the present invention lends itself to be implemented with a hardware structure (for example, integrated in chips of semiconductor material), or with a combination of software and hardware.

Claims

A method (400) for identifying signatures in a bookbinding machine including the steps of:
providing (404-418) a representation of each signature of a predetermined sequence to be repeatedly processed by the bookbinding machine,

for each cycle of the bookbinding machine involving the processing of a current signature, acquiring (404-405) a digital image of the current signature,

comparing (420-424) the digital image of the current signature with the representation of an expected signature in the sequence,
characterized by the steps of

in response to a negative result of the comparison of the digital image of the current signature, comparing (448-458) the digital image of the current signature with the representation of at least one signature being adjacent to the expected signature in the sequence, and

identifying (460) the current signature according to the results of the comparisons of the digital image of the current signature.
The method (400) according to claim 1, wherein the at least one adjacent signature consists of a signature preceding the expected signature in the sequence and of a signature following the expected signature in the sequence.
The method (400) according to claim 1 or 2, further including the steps of:
performing (434-440) at least one further cycle of the bookbinding machine,

comparing (464) a set of processed signatures including the current signature with a corresponding set of expected signatures, and

determining (465) an error condition of the bookbinding machine according to the result of the comparison of the processed signatures.
The method (400) according to claim 3, further including the steps of:
stopping (442) operation of the bookbinding machine, and

outputting (474) an indication of the error condition.
The method (400) according to claim 4, further including the steps of:
determining (472) a predefined correction action associated with the error condition, and

outputting (474) an indication of the correction action.
The method (400) according to claim 4 or 5, wherein the bookbinding machine consists of a sewing machine including an opening station being followed by a fixed saddle, a predefined number of cycles being required for bringing the current signature to the end of the opening station, and wherein the at least one further cycle of the bookbinding machine consists of said predefined number of cycles minus 1.
The method (400) according to any claim from 1 to 6, wherein the at least one further cycle of the bookbinding machine consists of two further cycles.
The method (400) according to any claim from 4 to 7, further including the steps of:
performing (468) at least one still further cycle of the bookbinding machine for facilitating an access to the current signature.
The method (400) according to any claim from 4 to 8, wherein the error condition is caused by at least one missing signature, the method further including the steps of:
performing (468) at least one idle cycle of the bookbinding machine to create room for inserting the at least one missing signature.
The method (400) according to any claim from 1 to 9, wherein the step of providing (404-418) the representation of each signature of the sequence includes, during a learning phase of the bookbinding machine:
acquiring (404-405) the digital image of a first signature in the sequence,

obtaining (411) the representation of the first signature from the corresponding digital image,

storing (412) the representation of the first signature, and

repeating the steps of:
acquiring (404-405) the digital image of a next signature in the sequence,

comparing (408-410) the digital image of the next signature with the representation of the first signature,

in response to a negative result of the comparison of the digital image of the next signature, obtaining (411) the representation of the next signature from the corresponding digital image and storing (412) the representation of the next signature,

otherwise, determining (414-418) the number of the signatures in the sequence according to the number of the next signatures and ending the learning phase.
A computer program (300) including program code means directly loadable into a working memory (230;265) of a data processing system (115,155) for performing the method of any claim from 1 to 10 when the program is run on the system.
A program product (235;265) including a computer readable medium embodying the program (300) of claim 11.
A bookbinding machine (100) including a data processing system (115;155) with a memory structure (235;270) storing the computer program (300) of claim 11 for carrying out the steps of the method according to any claim from 1 to 10.