WO1993020492A1 - Contour cutting system - Google Patents

Abstract

A contour cutting system comprises an electronic image detector (1) and a cutting device (5). Means (3, 7) are provided for processing signals from the image detector to drive the cutting device to cut a substrate in correspondence with an outline of an image detected by the image detector.

Description

CONTOUR CUTTING SYSTEM

The present invention relates to an automatic contour cutting system for cutting a substrate so that its outline corresponds to that of an observed image. One possible application in the cutting of silhouette pictures.

Silhouette pictures have been known for a long time. A dark substrate such as card is cut in profile to correspond to a silhouette picture of a sitter. The silhouette picture may then be mounted on a suitable background and/or in a frame. An example of such a silhouette picture is shown in Figure 1 of the accompanying drawings.

Until now, the cutting of silhouette pictures has not been done fully automatically from detection of the image to cutting of the substrate. It is an object of the present invention to provide a system of processing a captured image and using computer techniques to drive a cutting plotter over a substrate such as a piece of vinyl paper.

It is another object of the invention to provide various techniques for the enhancement of resolution of the captured image.

Thus the present invention provides a contour cutting system, said system comprising an electronic image detector, a cutting device and means for processing signals from the image detector to drive the cutting device to cut a profile in a substrate in correspondence with an outline of an image detected by the image detector. The Applicant may also claim a method of contour cutting, the method comprising viewing a subject with an electronic image detector and processing signals from the image detector to drive a cutting device to cut a substrate in correspondence with an outline of an image of the subject as viewed by the image detector.

One preferred application for systems according to the present invention is in the reproduction of silhouette pictures of people, for example as a machine situated in a retail outlet or public place.

The cut outline may, for example, correspond to a profile of the subject. For this purpose, the subject is preferably back-lit. In another alternative, the cut outline may correspond to a kind of portrait image of the subject. For that purpose, the subject is preferably side-lit or obliquely lit.

Preferably, the image detector is a solid-state imaging device such as of the charge-coupled device (CCD) type. It is also preferred to use an imaging device having a lower resolution (number of pixels) than the resolution (number of cutting steps of the cutting device).

Preferably, the system comprises electronic means for magnifying the image received by the imaging device to the order of size required by the cutting device, optionally with a facility for providing offsets (displacement) in the x and/or y orthogonal directions.

To overcome losses in profile sharpness, eg. resulting from magnification, the system preferably also is provided with electronic means for performing a veσtorisation routine. The electronic means may also be arranged to alter the received image to produce a stylised resultant image and/or to superimpose other features such as fashion accessories, on or adjacent the resultant image. Of course the system may be endowed with the capability to do both these things.

The substrate may be any suitable planar medium such as paper, card, vinyl paper or the like, the contour being cut out of the substrate. If the substrate is a hard material, the image contour may be produced as the edge bounding an etched area. Such etching may for example be performed by sand blasting or milling or grinding, eg. with a diamond tipped stylus or by any other suitable means.

If the substrate is a hard material to be etched, it may be substantially planar or in block form and may be any appropriate opaque, translucent or clear material, eg. glass.

The present invention will now be explained in more detail by the following description of a preferred embodiment and with reference to the accompanying drawings in which: -

Figure 1 shows a typical silhouette picture;

Figure 2 shows a block diagram of a system according to the present invention;

Figure 3 shows a set-up of the system shown in Figure 2, ready for use;

Figure 4 shows the sequence of electronic operations for the functioning of the system shown in Figure 2; Figure 5 is a diagram explaining the basis of a vectorization process;

Figure 6 shows the jagged outline resulting from use of an imperfect vectorization process;

Figure 7 is a diagram for explaining an improved three-point vectorization process utilised by the system shown in Figure 2;

Figure 8 is a diagram for explaining how the improved vectorization process deals with cutting a profile of the lips of a subject;

Figure 9 shows an arrangement of the system illustrated in Figure 2 for producing a portrait of a subject;

Figure 10 shows a typical portrait resulting from the arrangement shown in Figure 9;

Figure 11 and 11B show the result of a modified embodiment for producing sand-bladed images on glass blocks;

Figures 12A and 12B show the result of another modified embodiment for modifying the received image electronically to produce a stylised image; and

Figures 13A and 13B show the result of yet another modified embodiment for superimposing fashion accessories on the resultant image.

Some important features of the preferred embodiment include a motor driven mirror with CCD camera to mark position of the eyelashes. Special software helps the operator to mark position of the lips to obtain area where vectorisation would otherwise deteriorate the resolution of lip area where it is most critical for recognition. The applicant has undertaken significant development to be able to enhance the resolution of a 512 x 512 pixel CCD camera. Low resolution CCD cameras are preferred because they are relatively inexpensive, although higher resolution imaging devices may also be employed, eg. having 1024 x 1024 pixels.

The main components of the system of the preferred embodiment are shown in Figure 2. It comprises a solid-state imaging device in the form of a charge-coupled-device (CCD) camera 1 which provides electronic signals corresponding to a viewed image such as a person.

The signals from the CCD camera are processed by an image frame grabber 3 to drive a cutting machine 5 via a microcomputer and interface circuit 7. The image detected by the CCD camera is also displayed on a video monitor 9.

The image frame grabber 3 comprises an analogue-to-digital (D/A) converter or digitizer 11 which transfers the signals from the CCD camera to a frame memory 13 for storing data corresponding to image frames. The frame memory has one output 15 to the microcomputer interface circuit 7 and a second output 17 to display logic circuitry 19. The display logic circuitry supplies image signals to the video monitor 9.

The system is used as shown in Figure 3. A subject 21 is seated in front of a light source 23. The back-lit subject silhouette image is received by the CCD camera 1 and provides signals corresponding to an image frame to the microcomputer and interface circuit 7 via the image frame grabber (not shown in Figure 3). Under the control of software (to be described in more detail hereinbelow) the microcomputer drives the cutting machine via the interface circuit to cut a substrate in the form of the image profile.

A general overview of the system of the preferred embodiment will now be described.

The machine is designed for retail sales of instant silhouette images of subject customers, eg. in souvenir shops on tourist sites. The customer will be able to use the equipment in a booth were he or she will be seated on a seat provided. The customer will look at a video monitor on which his or her own profile will be displayed. The machine operator will position the camera through a joystick to centralise the image.

When both parties are satisfied with the image either one of them would press a button to initiate the cut. The image will be cut by a cutting plotter, preferably on monochrome self adhesive vinyl paper. The cut image can then be lifted off and pasted onto a printed postcard with appropriate background sceneries. A stored program of various favourite animal images and horoscope signs and other attractions are available in a software library. These may be called by the program and superimposed on the subject and edited for variations in presentation. Examples of the latter will be explained in more detail hereinbelow.

Along with the cut images, the operator can also provide picture frames, postcards, special attractions cards, local scene cards, epoxy casts and the like.

The embodiment use a cutting plotter sold under the a KANTO Machinery Trade Mark PREART ex Kanto Machinery Corp. of Japan. This cutter has its own processor and ROM subroutine and buffer memories. It takes position commands from the system software and and makes its own decisions during acute turns to accommodate the swivel of its offset centre blade.

As mentioned above, the preferred CCD camera has a 512 x 512 pixel array resolution device. The Kanto cutter has a resolution of 3600 steps across a width of six inches of vinyl paper. The length of the cut piece is only limited by the system software. Obviously, a one-to-one transformation would only use one twelfth of the paper and produce a cut image of the size of a postage stamp. To overcome this problem the system has a magnification factor written into the software. This is able to transform the coordinates onto a 3600 x 3600 array, and then add in an offset constant and position the image at the centre of the paper. In other words, mathematically each coordinate is expressed thus:

X coordinate = MIX + Cl Y coordinate = M2Y + C2

where Cl and C2 are offset constants and Ml and M2 are magnification factors (Note that the image at this stage has been transformed from line (raster) to chaincode coordinates). However, this creates another problem in that the new image has a rather fuzzy outline and one can actually see the discreet steps from one pixel to the next. Moreover, it also takes the cutter too much time to follow and make acute 90 degrees turns on each step. To overcome this problem, "vectorisation" routine has been included to even out the steps. The basic vectorisation routine may be understood by reference to Figure 5. A straight line other than in the x or y axis direction or at 45^* to the x and y axes will be mapped as steps when it is down to single pixel resolution.

However, this leaves the image somewhat compromised in details. The result is still not optional as the sharp incut of the space between the upper and lower lips disappears along with the vectorised data. The applicant has found that the sharp detail of the area around the lips is most important for accurate resemblance and hence personal recognition. When the steps are mapped onto the cutter software, they are hugely magnified and the picture would be represented as visilbe steps. The result would be an image similar to the one represented in Figure 6.

In order to avoid such data loss the basic vectorisation routine was modified as follows to avoid compromising around the lips area.

The Applicant has^' devised a "3-point average" vectorisation routine to smooth-out the steps. This routine takes 3 consecutive points in the image, finds an average and assigns this as the vectorisation point.

For example, as shown in Figure 7, if A, B, C, D, E, F and G are the original points appearing in steps:

Point X.=average of A, B & C

X,=average of B, C & D

X.=average of C, D & E and so on.

Thus the vectorised points are ., X,, X,, X., X₅ etc. Therefore, the cutter will trace through the X points and so the visible steps are effectively eliminated. The way in which this routine deals with the area around the lips may be understood by reference to Figure 8. In this drawing, the circles trace the original curve. The X' s trace the vectorised path. Reference numeral 25 denotes the upper lip, 27 the lower lip and 31, the chin. It is found that the lips disappear as a line across the upper and lower lips.

In order to overcome this problem, the routine marks the point between the upper and lower lips and instructs by means of the software that 5 pixels above and below that point "do not take average" that is to say, present it as it is. This has proved particularly successful in overcoming the aforementioned problem.

Instead of cutting an outline corresponding to a profile of a subject, the system of the present invention may be used to cut an outline corresponding to the boundary between light and dark areas of a side-illuminated of obliquely-illuminated subject. This results in an aesthetically pleasing cut image of the subject, as if painted by an artist.

The arrangement for producing such a cut image is shown in Figure 9. The subject 21 is seated in front of a white board 33 and illuminated obliquely by the light source 23. The CCD camera 1 is located in front of the subject to view both the light (illuminated) part 35 and the dark (in shadow) part 37. After contour cutting, the resultant image is as shown in Figure 10.

The embodiments described above cuts the outline from a planar blank such as vinyl paper. In an alternative embodiment, the outline may be produced by sand-blasting an area of a solid substrate, in this case a glass block. Examples are shown in Figures 11A and 11B. It can be seen that appropriate lighting of the subject also enables reproduction of image detail within the boundary of the overall profile contour, as with the image shown in Figure 10.

Figures 12A and 12B show how embellishments to the basic image may be made to enhance the appearance of the subject. Figure 12A shows the basic image which would be cut, eg. on vinyl paper. As shown in Figure 12B, the system software may be used to reverse the image, to add (for example) hair highlights 43, new hairstyle 45, hat 47 and adjacent decorative features 49.

In general, the software may be used to superimpose accessories on an image as shown in Figures 13A (on sandblasted glass) and 13B (cut paper contour). These may (for example) be headbands 51, 53, 55 and ear-rings 57, 59.

The order of operational steps of the system is controlled by the microcomputer under the control of a software algorithm, the main steps of which will now be described with particular reference to Figure 4. These steps are: -

A. Image Grabbing

B. Edge Smoothing

C. Thresholding

D. Line Extracting

E. Data Linking

F. Raster to Vector Conversion

G. Plotting and Cutting

Details of these steps are as follows: - A. Image Grabbing Process

A physical object or pattern is photographed by using the CCD camera. The analogue image grabbed by the camera is digitized into a digital image which are then stored in the frame memory as an original pattern image.

B. Image Smoothing Process

Occasionally, a grabbed image may contain some noises. Therefore, after capturing the image, the image signals are then "smoothed" to remove noise and thus helping the process of thresholding described in the following step.

C. Image Thresholding Process

The purpose of the system is to extract the profile or contour of the object or pattern and to transform the raster data to vectors; in other words, the system deals with lines (profile or contour) only. In order to reduce the data and to simplify the subsequent processing, the original pattern image is first thresholded into a binary image which is classified into two parts, the subject and the background.

D. Line Extranting Process

After the original pattern image is thresholded, the binarized image is then processed to extract the outline of the image. This is done by scanning the pattern image stored in the frame memory from left to right and from top to bottom. E. Data Linking Process

After the process of line extraction, the original image is then reduced into some lines (profile or contour) but these data still consist of discrete raster points. It is convenient to store the points before the process of raster to vector conversion proceeds.

F. Data Conversion Process

The vectorization technique used is chain encoding. After a point is is vectorized, the grey level of that point is changed to the same value as that of the background in order to erase it and prevent from being redundantly searched. The vector point which is the direction code of the neighbouring pixel is converted to a normal x,y coordinate. This is done because the cutting machine moves or performs its plotting or cutting functions with respect to normal x,y-σoordinates.

Therefore, instead of storing direction codes, the system will store x,y-σoordinates. As a result, the system will store all the coordinates sequentially in the same order as they were searched.

G. PIotting/Cutting

Once all the x,y-coordinates of the image profile are saved, the profile can be plotted or cut by the cutting machine.

Specifies of a particular commercial embodiment will now be described in detail, together with details of the εoftware. The PROFILE CUTTING SYSTEM (PCS) is a based on a software development HRT-512-8 Video Frame Grabber library function. The software development of the PCS is written in Microsoft C, version 6.0. However it will be appreciated that the system may be realised in any appropriate computer language, high or low level.

The main objective was to develop a vision system which functions as the ' eye' of the UCS.

This system must be able to capture a live image, process the image to obtain its profile and finally cut the profile with the specified cutting machine.

The PROFILE CUTTING SYSTEM uses: -

User-friendly menu driven software window Image processing Vectorisation process Profile cutting

The Equipment used is: -

IBM PC/AT or 100% compatible equipment HRT-512-8 Video Frame Grabber Video Monitor PHILIPS CCD camera COMPUTAR TV lens - 25mm KAISER RSI camera rack PREART Computer Controlled Cutting System

(a) HARDWARE INSTALLATION

The HRT-512-8 card is shipped with the factory setting of OD400H for the base segment address of the digitizer. This are the default setting that the PCS software is set up for. The connectors used are as follows:

Top RCA jack is for NTSCPAL video input. Bottom RCA jack is for video output (monitor). Top dial is for video offset (brightness). Bottom dial is for video gain (contrast).

(b) SOFTWARE INSTALLATION

To ensure correct operation of the PCS, two software packages must be installed properly. They are: Microsoft C Computer, and the PCS software package. The packages are to be installed using the defaults provided by their associated installation or set-up programs. The Microsoft C compiler version 6.0 is installed using the set-up program and accepting all the defaults. To complete the installation of the C compiler and the HRT libraries, the AUTOEXEC.BAT and CONFIG.SYS must be supplemented. Within the AUTOEXEC. BAT file, the amendments are as follows:

PATH = C: \C600\BINB;C: \C600\BIN; SET-LIB = C:\C6001\LIB; SET INCLUDE = C: \C6001\INCLUDE; SET HELPFILES = C: \C600\HELP\*. HLP; SET INIT = C: \C6001\INIT;

If the user already has directories listed in any of the preσeeding commands, those specified are simply to the end of the user list.

Within the CONFIG. SYS file, the following statement should be included:

DEVICE = C: \C6001\BIN\HIMEM. SYS If the user is still not familiar with the following changes, a sample entry of how to implement AUTOEXEC. BAT is found in the NEW-VARS. BAT file whereas CONFIG. SYS is found in the NEW-CONF. SYS file that SETUP creates. Note that both these files (AUTOEXEC.BAT and CONFIG.SYS) should be included in the root directory of the disk from which the user boot, normally the hard disk drive (C) Before installing the PCS Software Package, it must be ensured that the directory and sub-directories of the PCS disk are created in the root directory. Once created, the PCS disk is placed into drive A: all the files are copied into their respective directories and sub-directories by typing:

COPY A: \PCS\DATA\*. * C: \PCS\DATA COPY A: \PCS\PONT\*. * C: \PCS\FONT COPY A: \PCS\LIB\*. * C: \PCS\LIB COPY A: \PCS\*. * C: \PCS

It is recommended that for a better compilation speed, the MSHRT. LIB file should be copied to the C compiler by typing:

COPY C: \PCS\LIB\*. * C: \C6001\LIB

For compiling and linking, all the source, header and the object files are included. For convenience, the object files are precompiled under the C small model.

A "MAKE" file called PCS is provided solely for Microsoft C version 6.0. These files are to be used with the "MAKE" utility program supplied with the Microsoft C compiler. If the utility program displayed errors, the "MAKE" file that is provided by the PCS software is used. This file (PCS) will regenerate the object modules the PCS. EXE program if any changes have been made to source code. To regenerate the PCS. EXE program after source code has been modified, the following command typed: -

MAKE PCS

The Microsoft utility will replace any object files which are no longer current and relink them together to produce the new executable program. Once a source code file has been created and modified, both the make file and the link file have to be modified accordingly. Within the make file (PCS), the editor is used to add another two lines like those already present to invoke the Microsoft-compiler. An example for C version 6.0 is shown below:

NEWFILE. OBJ : NEWILE. C CL $(C_SET) $**

NEWFILE. OBJ must then be added to the list after PCS. EXE near the bottom of the make file. The link file (PCS. LNK) is modified by adding the line:

NEWFILE +

to the existing list. If it is desired to change the C model size, it is necessary build the appropriate library using Microsoft' s Library Manager and change the supplied make file to reflect the chosen model

The overall structure of the software is as follows: -

- MISC

- SCREEN

- VIEW

- TRACE - CUT

- EYE

- BASE

- FRAME

PROGRAM CODE

EXECUTION CODE

PCS. -EXE

COMPILATION CODE

PCS

PCS. -LINK

MME. -EXE

HEADER FILES

PCS-C & PCS. OBJ MISC-C & MISC. OBJ SCREEN. C & SCREEN. OBJ VIEW. C & VIEW. OBJ TRACE. C & TRACE. OBJ CUT. C & CUT. OBJ EYE. C & EYE. OBJ BASE. C & BASE. OBJ FRAME. C & FRAME. OBJ The software modules are as follows: -

File: PCS

The file contains the routine that generates the main menu. It also generates help messages regarding the available options present in the main menu.

File:MISC

This file contains miscellaneous routines to supplement the main menu, such as freeing memory spaces allocated during the execution of the software, time delay and clearing of the menu display. It also contains routines that create graphics cursor, selecting graphics font and displaying characters keyed-in from the keyboard in the selected graphics font.

File: SCREEN

This file contains routines that display the live image on the video monitor and clearing the video monitor' s screen. It also contains routine to draw a cursor on the video monitor which is used to locate the position of where to add the eye-lashes.

File:VIEW

This file is the foundation of the vision system. It contains routines to capture and process the image according to the requirements of these system.

File: TRACE This file contains a routine to trace the profile of the image. It also calls on another module EYE which adds eye lashes to the image profile if required. File: CUT

This file contains a routine to produce a hardcopy of the image profile. It also calls on module BASE which add a' sculpture base' to the image profile, and module FRAME which add a frame to the image profile.

File: EYE

This file contains routine to add eye lashes to the image profile. This file is called by module TRACE.

File: BASE

This file contains routine to add a ' sculpture base' to the image profile. This file is called by module CUT.

File FRAME

This file contains routine to add a frame to the image profile. This file is called by module CUT.

The main software system parameters will now be described in more detail: -

HOST SYSTEM /

The system memory must be large enough for the allocation of memory spaces. The size of the system hard disk must also be large enough to cater for the storing of the image files, it is recommended that all image files (extension . tif) be saved in a Floppy disk whereas all image data files (extension . dat and extension . σde) present in the directory C: \PCS be deleted so as not to waste the memory space of the hard disk. GRAPHICS ADAPTER

The graphics adapter of the host system should be VGA or better since both the software package and the frame grabber used required this type of graphics adapter.

System will not operate if the VGA card is absent since both software package and video frame grabber card require a VGA card to operate.

CAMERA AND LENS

The camera and lens must be regularly clean of dust since it will affect the image processing. When system is not in use the camera lens should preferably be covered.

HRT-512-8 VIDEO FRAME GRABBER

The video input and video output of the frame grabber card must be correctly connected. These relevant connectors should be secured to the RCA jacks if possible. It is also preferred that the frame grabber card be tested to check whether it is functioning properly by running the software package that comes together with the purchase of the frame grabber.

In the light of this disclosure, modifications of the preferred embodiment as well as other embodiments all within the scope of the appended claims will become apparent to those skilled in the art.

Claims

CLAIMS: -

1. A contour cutting system, said system comprising an electronic image detector, a cutting device and means for processing signals from the image detector to drive the cutting device to cut a substrate in correspondence with an outline of an image detected by the image detector.

2. A system according to claim 1, wherein the image detector is a solid-state imaging device.

3. A system according to either preceding claim wherein the image detector has a lower detection resolution than the cutting resolution of the cutting device.

4. A system according to claim 3, wherein the signal processing means comprises means for electronically magnifying the image.

5. A system according to claim 4, wherein the signal processing means comprises means for electronically effecting a vectorization process on the magnified image signals.

6. A system according to any preceding claim, adapted for cutting a silhouette picture of a person.

7. A system according to any of claims 1-5, adapted for making a portrait.

8. A system according to claim 7, wherein the portrait is a profile portrait.

9. A system according to claim 7, wherein the portrait is a front view.

10. A system according to any of claims 6-9, wherein fashion accessories may be superimposed on the resultant image.

11. A system according to any of claim 6-10, wherein the image received by the image detector may be electronically modified to produce a stylised resultant image.

12. A system according to any preceding claim, adapted to cut a contour in a planar substrate.

13. A system according to any preceding claim, adapted to etch an area on a substrate.

14. A system according to claim 13, wherein the substrate is in block form.

15. A system according to claim 14, wherein the substrate is a glass block.