MXPA98001012A - Bottle thread inspection system and method to operate the mi - Google Patents

Abstract

The present invention relates to a method for inspecting a bottle (14), having a threaded section (104), for thread defects as the bottle moves along a production line (810), which requires that video images (100) be taken from the bottle as the bottle moves towards the fields of vision of video cameras (22). Each video image encompasses a general region of interest (108), which contains at least a portion of the threaded section of the bottle. The position of the bottle (14) within the video image is then determined based on the location of the upper flange (110) of the bottle within the video image. A portion of the general region of interest, which comprises at least a portion of the threaded section of the bottle, is segmented to a plurality of specific regions of interest (120, 122). The pixels of the video image, from the specific regions of interest, are then examined to detect defects in the rosc

Description

BOTTLE THREAD INSPECTION SYSTEM AND METHOD TO OPERATE THE SAME FIELD OF THE INVENTION The present invention relates to bottle inspection systems and in particular to an inspection system and method for detecting bottleneck defects.

BACKGROUND OF THE INVENTION Conventional bottle processing equipment moves the bottles at a high speed along a production line. It is important to inspect the bottles to detect bottles that have defects. These defects include, but are not limited to, cracks or burrs within the area of the thread around the neck of the bottle. When necessary, bottles with defects are rejected, that is, removed from the production line. Due to the speed at which the production lines operate, a very rapid defect detection is required. Computerized video image analysis is well suited for this purpose, due to its contactless nature, high speed, ability to make decisions, and ability to analyze a large area of bottles with each video image. Inspection systems have been developed to detect bottleneck defects in bottles moving along a production line. For example, the patent of E. U.A. No. 3,848,742 to Krenmayr describes a system for detecting various types of defects within the glass in the neck of a bottle. During the operation of this system, the bottle is rotated in place 360 degrees around its central vertical axis and must be accurately positioned in the inspection area. These requirements reduce the inspection speed of the system by allowing the system to inspect only a small number of bottles per minute (for example, 200 bottles of 90 g per minute). The patent of E. U.A. Do not . 4, 701, 612 of Strurgil and the patent of E.U.A. No. 4, 958,223 of J uvinall, both have the same assignee, describe inspection systems, which require that glass or plastic containers be maintained in vertical orientations and rotated 360 degrees around their central axes. In addition, the systems require a gray level pattern comparison, that is, the gray level image of the area of the bottle under inspection must be compared with a normal image indicative of an acceptable container. To implement pattern comparison of this nature, pre-processing is required to achieve normal orientation before performing the pattern comparison procedure. All these requirements result in a deficiency of the total low inspection regime. The E. U .A. No. 5, 126, 556 of Do enico and others describes three methods for inspecting bottles to detect defects in the threads. The first two methods are based on the precise placement of a bottle in the inspection area. Specifically, the central vertical axis of a bottle must coincide with the optical axis of the image formation system. The third method, however, requires that a bottle be rotated 90 degrees around its central vertical axis, while the bottle is moving along a conveyor. In these three methods, defect detection is based on a pattern comparison at the gray level. As mentioned above, pre-processing is required before the pattern comparison can be implemented. The patent of E. U.A. No. 5,444, 535 of Axelrod discloses a high signal to noise optical apparatus and a method for detecting damage to the thread of a glass bottle. The apparatus includes a source for directing light against a target glass surface. The source is selected to emit light at wavelengths substantially overlapping a target glass absorption bandwidth. A first optical polarizer polarizes the light emitted from the light source before the light hits the surface of objective glass. A light detector in the form of a photodetector and a second optical polarizer disposed in a polarized relationship transverse to the first optical polarizer are aligned in a dissipated light beam detection ratio relative to the incident light beam on the target surface of the light source. glass. The light detector is at an angle on the Brewster scale and generates a detected signal in response to the light dissipated through the defects in the target glass surface. Although the above references describe inspection systems for detecting bottleneck defects, improved systems are continually being sought to detect bottleneck defects more quickly and more accurately.

Therefore, it is an object of the present invention to provide an improved method and system for inspecting bottles to detect thread defects.

COMPENDIUM OF THE INVENTION In accordance with one aspect of the present invention, there is provided a method for inspecting a bottle having a threaded section for thread defects as said bottle moves along a production line, comprising the steps of: (i) capturing a video image of said bottle with a video camera as the bottle moves into the field of view of said video camera without requiring that said bottle be in a specific position within the field of view, said video image comprising a general region of interest containing at least a portion of the threaded section of said bottle; (ii) determining the position of the bottle within the general region of interest based on the location of a feature of said bottle within the general region of interest; (iii) segmenting a portion of said general region of interest, to which it encompasses at least a portion of the threaded section in a plurality of specific regions of interest; and (iv) examining the pixels of said video image in the specific regions of interest to detect thread defects. In the preferred embodiment, the feature of the bottle determined in step (ii) is the top flange of the bottle. Once the upper flange of the bottle is established, then a reference location is determined based on the position of the bottle. The portion of the general region of interest and the positions of the specific regions of interest are then determined based on the reference location position. In one embodiment, the reference location is positioned in the center of the upper flange and a specific central region of interest is determined relative to the reference location. The positions of the other specific regions of interest on the opposite sides of the specific central region of interest are then determined in relation to the position of the specific central region of interest. Preferably, the other specific regions of interest are greatly deviated in a Y direction, whereby they are formed from the specific central region of interest to follow the threaded section of the bottle in the video image. It is also preferred that the other specific regions of interest reduce their width (i.e., in an X direction), so that they are from the specific central region of interest to compensate for effects in perspective. In a preferred embodiment, during step (iv), a black / white pixel threshold is determined for the specific regions of interest. The pixels of the video image within the specific regions of interest are compared with the threshold value of the pixel and the pixels are placed as binary, such as white or black, depending on the results of the comparisons. Contiguous white pixel groups greater than a threshold number are filtered, and if the shadows of adjacent white pixel groups do not resemble bottle threads, groups of contiguous white pixels are determined as thread defects. According to another aspect of the present invention, there is provided a system for inspecting bottles having a threaded section as the bottles move along a production line, comprising: a plurality of image forming sections of video, arranged along the production line in separate locations, each video image forming section being oriented with respect to said production line to take a video image of each bottle in a different circumferential region thereof, to As each bottle moves into the field of view of the video image forming section without requiring the bottle to be at a specific location in the field of view, said video images encompassing a general region of interest that contains minus a portion of the threaded section of each bottle; and processing means in communication with the video image forming sections and receiving the video images taken by them, said processing means processing each video image to determine the position of the bottle within the general region of interest based on the location of a feature of said bottle within the general region of interest; segmenting a portion of the general region of interest, which encompasses at least a portion of the threaded section to a plurality of specific regions of interest; and examining the pixels in the specific regions of interest to detect thread defects. In a preferred embodiment, each video image forming section includes a video camera and a light source. The light source and the video camera are placed on opposite sides of the production line and are oriented so that video images of the entire circumference of each bottle are taken. In one embodiment, the processor means signal a bottle reject mechanism downstream of the inspection system when a defective bottle is detected, so that the defective bottle can be removed from the production line. According to yet another aspect of the present invention, there is provided a system for inspecting bottles having a threaded section as said bottles move along a production line, which comprises: a plurality of image forming sections of video arranged along said production line in separate locations, each video image forming section being oriented with respect to said production line to take a video image of each bottle in a different circumferential region thereof, a As each bottle moves into the field of view of the video image forming section without requiring the bottles to be in a specific location in the field of view, said video image forming sections are arranged to reduce the spacing between them.; and processing means in communication with the video image forming sections and receiving the video images taken by them, said processing means processing the video images to detect thread defects in said bottle. In a preferred embodiment, each video image forming section includes a video camera and a light source. The light source and the video camera are positioned on opposite sides of the production line and are laterally deflected, so that the optical axes of the video image forming sections form oblique angles with respect to the direction of travel of the images. bottles It is also preferred that the video image forming sections be arranged in an upstream pair and a downstream pair with the optical axes of the upstream pair forming angles with respect to the path direction of the bottles, and with the optical axes of the Downstream torque forms acute angles with respect to the direction of trajectory of the bottles. According to a further aspect of the present invention, there is provided a method for locating the position of a bottle within a video image, comprising the steps of: capturing a video image of at least a portion of said bottle with A video camera; digitize said video image; comparing the pixels in the video image with a threshold value and biasing said pixels as white or black depending on the results of said comparisons; and determining the position of said bottle relative to the boundaries of the video image based on the location of the largest group of contiguous white pixels within said video image. In a further aspect of the present invention, a method for detecting defects in bottle threads is provided, which comprises the steps of: capturing a video image of said bottle threads; segmenting at least a portion of the video image to a plurality of specific regions of interest; and process the video of each specific region of interest, independently, to determine defects in each of the specific regions of interest, based on the existence of white areas greater than a prescribed area size threshold within said specific regions of interest .

According to a further aspect of the present invention, there is provided a system for locating the position of a bottle within a video image, which comprises: means for illuminating said bottle with a light source against light; a video camera for capturing a video image of at least a portion of the bottle within the field of view thereof; means for digitizing said video image; and means for determining the position of the bottle within the video image, based on the location of the largest white area within said video image. According to a further aspect of the present invention, there is provided a system for detecting defects in bottle threads, which comprises: means for capturing a video image of the bottle threads; means for segmenting at least a portion of the video image to a plurality of specific regions of interest; and video processing means of each specific region of interest, independently, to determine the defects in each of the specific regions of interest based on the existence of white areas greater than a prescribed area size threshold within the specific region of interest. interest. The present invention provides advantages since screw defects in bottles can be detected without requiring the handling of the bottles and while the bottles are moving at a high speed along the production line. When a defective bottle is detected, the inspection system signals a bottle reject mechanism to allow the defective bottle to be removed from the production line without reducing the movement of the bottles along the production line.

BRIEF DESCRIPTION OF THE DRAWINGS The embodiments of the present invention will now be described more fully with reference to the accompanying drawings, in which: Figure 1 is a schematic side elevation view of a portion of a bottle production line including an inspection system to detect defects in bottles, according to the present invention; Figure 2 is a schematic block diagram of the inspection system illustrated in Figure 1; Figure 3 is a top plan view of a portion of the inspection system illustrated in Figure 1; Figure 4 is a side elevational view of a portion of the inspection system illustrated in Figure 1; Figure 5 is an illustration of a video image of a bottle, taken by the inspection system of Figure 1; Figure 6 is an illustration of a binary video image and showing specific regions of interest within the video image; Figure 7a is an illustration of the specific regions of interest within the video image after being binary to show thread defects; Figure 7b is an illustration of the specific regions of interest containing the significant thread defects; Figure 8 is an illustration of a binary video image and showing an alternative embodiment of the specific regions of interest within the video image; Figure 9 is an illustration of specific regions of interest grouped shown in Figure 8; Figure 10 is a plan view of an alternative embodiment of a bottle production line including an inspection system for detecting bottle defects; Figure 11 is a schematic block diagram of the inspection system illustrated in Figure 10; Figure 12 is a perspective view showing a portion of the inspection system illustrated in Figure 1 1; and Figure 13 is an illustration of a binary video image taken by the inspection system of Figure 10, showing the specific regions of interest and the thread defects therein.

DESCRIPTION OF THE PREFERRED MODALITIES Referring now to Figure 1, a potion of a line is shown in a bottling facility, and is generally indicated with the reference number 10. The production line includes a conveyor system 12 for moving glass bottles 14 along a path indicated with the number of reference 15 to a high speed regime typically in the range of 800 to 1350 bottles per minute between several stations located along the production line. Positioned along the production line is an inspection system 16 for inspecting the threaded section of each glass bottle 14 for defects as the bottles move along the conveyor system 12 without the bottles being handled. Downstream of the inspection system 16 is a bottle reject mechanism 18. The bottle reject mechanism 18 responds to the inspection system 16 and removes the defective bottles from the production line, so that these bottles do not travel downstream to other stations in the installation in botelladora. Referring now to Figure 2, the inspection system 16 is better illustrated. As seen, the inspection system includes a plurality of video image forming sections 20 positioned above the conveyor system 12 at separate locations. Each video image forming section 20 includes a 22 Pulnix CCD camera and a halogen lamp 24. A bottle detection sensor 26 is associated with each video image forming section 20. The bottle detection sensors are positioned at along the conveyor system 12 adjacent to its associated video image forming section 20. An additional bottle detection sensor 28 is positioned along the conveyor system 12 downstream of the video image forming sections 20 to detect bottles as they exit the inspection system 16. The bottle detection sensors 26 and preferably they are photoelectric, such as those manufactured by Omron. The bottle sensing sensors 26 and 28, the CCD cameras 22 and the halogen lamps 24 are connected to a junction board 30. A temperature sensor 32, an encoder 34 and a lamp power supply 36 are also present. connected to the junction board 30. The encoder 34 is located along the conveyor system 12 and detects the speed of the bottles 14 as they move along the conveyor system 12. The junction board 30 is also connected to a alarm 38, in the form of a guide, and a bottle rejection mechanism 18. Also connected to the junction board 30 is a computer 40.

The computer 40 includes a plurality of video image processors 42, each video image processor of which is associated with one of the CCD cameras 22. Each video image processor 42 includes a secondary ARTVC card for capturing video images. video output from the CCD cameras 22 and a motherboard to process and analyze the video images captured by the secondary card to detect defects in bottle threads, as will be described. The video image processors 42 capture and process video images that are output from the CCD cameras 22 in response to the bottle detection sensors 26. An online production board 44 within the computer 40 also responds to the detection sensors of bottles 26 and 28 and to the coded 34 and maintains the lane of bottle positions as they move through the inspection system 16 to allow detected defective bottles to be tracked and removed from the conveyor system 12 through the mechanism bottle rejection 18. A PCOM card 46 within the computer 40 verifies and tests the inspection system 16 to detect the alconditions, as will be described. An operator monitor 48 and an associated touch screen 50 are also connected to the computer 40 to allow an operator to enter commands to the inspection system 16 and see the results of the inspections of the bottles. A live monitor 52 is also connected to the computer 40 through the junction board 30 and displays a video image of the last defective bottle to pass to a video image forming section 20. Input devices, such as a mouse 54 (mouse) and a keyboard 56 are also connected to the computer 40 to allow the inspection system to be initialized and reprogrammed if required.

The components of the inspection system 16, with the exception of the bottle detection sensors 26 and 28, are adapted in a cabinet 60, which straddles the conveyor system 12. Fans 62 are mounted in the cabinet 60 to moderate the temperature in the same. A switchable power supply module 64 is connected to AC conductors and supplies power to the fans 62, the lamp power supply 54, the live and operator monitors 52 and 48, respectively, and the computer 40. Referring to the Figure 3, the video forming sections 20 are better illustrated. As can be seen, the CCD camera 22 and the halogen lamp 24 of each video image forming section are placed on opposite sides of the conveyor system 12. The CC cameras D and the halogen lamps are also arranged in protected pairs, upstream and downstream, locked, 66 and 68, respectively to reduce the length of the inspection system 16 along the conveyor system 12 and to reduce the interference between the forming sections adjacent video image 20. Specifically, the CCD camera 22 and the halogen lamp 24 in each of the video image forming sections or 20 are positioned on opposite sides of the conveyor system 12 and are laterally offset so that the OA of the optical axis of each video image forming section forms an oblique angle with respect to the path 15 of the bottles, as they travel along the conveyor system 12. In particular, the optical axes of the upstream pair 66 of the video image forming sections 20 form obtuse angles with respect to the path 15, while the optical axes of the downstream pair 68 of the video image forming sections 20 form acute angles with respect to the path 15. The positions of the CCD cameras 22 and the halogen lamps 24 with respect to the sides of the conveyor system 12 are alternated in the video image forming sections 20. successive The field of view of each CCD camera 22 includes approximately 1 10 degrees of the circumference of the threaded section of a bottle 14. The alternating positions of the CCD cameras 22 and the halogen lamps 24 in the successive video image forming sections 20 and the orientation of the CC D cameras and the halogen lamps in the video image forming sections 20 ensure that the fields of view of the CCD cameras 22 overlap and encompass the entire circumference of a bottle 14 as it travels through it. of the inspection system 16. The reduced space between the successive video image forming sections 20 and the overlapping fields of view of the CCD cameras 22 reduce to a minimum the probability of the rotation of a bottle on the conveyor system 12 around its central longitudinal axis, as it travels between the successive video image forming sections by a quantity, which results in a region of the threaded section of the bottle that is not captured in a video image taken by any of the image forming sections of the bottle. video 20. In this way, multiple video images of the bottle can be taken asynchronously through the video image forming sections 20, while ensuring that the video images encompass the entire circumference of the threaded section of the bottle . Figure 4 illustrates much better the orientation of the CCD camera 22 and the halogen lamp 24, in one of the video image forming sections 20. Those skilled in the art will appreciate that the orientation of the CCD cameras 22 and the lights of the halogen 24 is the same in each of the video image forming sections. As can be seen, the halogen lamp 24 is suspended from the cabinet 60 and oriented to direct light 70 downward toward a bottle 14 at an angle of about 45 degrees to illuminate the bottle against light. A double glazed window 72 is placed in front of the halogen lamp 24 and is oriented so that the light beam 70 is generally normal to the plane of the glass window 72. One of the window panes 72 is slidable with relationship to the other of the crystals, to allow the window 72 to be cleaned. The CCD camera 22 is suspended from the cabinet 60 and is oriented with its lens 74 pointing in a direction parallel to the longitudinal axis of the bottle 14. An angled mirror 76, placed below the CCD camera 22, is also suspended from the cabinet and directs the reflected light from the bottle 14 to the CCD camera 22. The mirror is angled so as to form an angle of approximately 65 degrees with respect to the plane of the lens 74 of the CCD camera 22. A double-glazed window 78 is placed between the mirror 76 and the bottle 1 and is oriented so that its plane is normal to the light reflected from the bottle and directed towards the mirror 76. One of the crystals of the window 78 is slidable relative to the other of the crystals, to allow window 78 to be cleaned. The windows 72 and 78 can be coated, if desired, with a coating against reflection. It has been found that the angled direction of light 70 generated by the halogen lamp 24 and the orientation of the mirror 76 and the windows 72 and 78 allow the defects to be consistently detected, while the total internal reflection is reduced. Although not illustrated, the windows 72 and 78 are mounted on an inverted zigzag-shaped support extending through the cabinet 60. This allows the support to adapt the windows 72 and 78 of each video image forming section. in separate locations along its length. During operation, when a bottle 14 moves along the conveyor system 12 and passes to a bottle detection sensor 26, the bottle detection sensor 26 detects the presence of the bottle 14 and outputs a detection signal of the bottle. bottle. The bottle detection signal is received through the junction board 30 and passes to the associated CCD camera 22 as well as to the computer 40. The CCD camera 22 responds to the bottle detection signal by opening its shuttle, so that a Video image 100 is taken from the rim 102 and the threads 104 of the bottle 14, as shown in Figure 5. The computer 40 uses the bottle detection signal to activate the video image processor 42 associated with the CCD camera 22 so that the video image taken by the CCD camera 22 is captured by the secondary card therein. The computer 40 also transports the bottle detection signal to the production line board 44, so that the bottle 14 can be counted and its position, as it travels through the inspection system 16., tracked. The video image 100 captured by the sub card in the video image processor 42 is then analyzed by the CPU in the video image processor 42 motherboard to detect thread defects, as will be described. Since the inspection system 16 includes four bottle detection sensors 26 and associated CCD cameras 22 and video image processing 42, the above procedure is performed four times, as each bottle travels through the inspection system 16. If no significant defects are found in the threads 194 of the bottle 14, as the video images 100 are analyzed, the bottle is allowed to continue along the conveyor system 12. However, if a significant defect is detected in the bottle. the threads 104 of the bottle 14, the production line board 44 sends a signal to the bottle reject mechanism 18 via the junction board 30 to allow the bottle reject mechanism to remove the defective bottle from the conveyor system 12 when the defective bottle arrives at the bottle rejection mechanism. Specific points will now be described in the way in which the video images captured by the CPUs in the motherboards of the video image processor are analyzed to detect defects in the bottle thread. When a video image 100 is captured through a video image processor 42, the video image is digitized and stored in memory as a two-dimensional array of pixels, typically 512 x 480 pixels, covering a field of view of approximately 50 x 50 mm, so that each pixel represents approximately 0.1 mm x 0.1 mm of the video image. After this, the CP U on the motherboard collects a histogram on the pixels in a rectangular window 108 within the video image 100. The rectangular window 108 may include between 75% and 100% of the pixels within the image video 100. Then, a black / white threshold heat is calculated for the video source 100, using the Otsu method, as described in an article entitled "A Threshold Selection Method From Gray-level Histogram" by H. Otsu, in I E EE Transactions on Systems, Man and Cybernetics, Vol. SMC-9, page 62-69, 1979, the content of which is incorporated herein by reference.

The calculated threshold value is then diverted by an empirically determined amount. The actual threshold value used is based on an average of operation of the deviated Otsu threshold values calculated above for the video images captured by the video image capture card 42. This allows the inspection system 16 to compensate for aging, accumulation of dust, etc. , which typically results in more obscure video images. The pixels in each scan line within the window 108 are then encoded in the path length. In particular, initially the pixels in each scan line are assumed to have a gray level value below the actual threshold value. The adjacent pixels in each scan line are then compared. During this comparison, the directions of the pixels, which change from the low real threshold value to a real threshold value above and which change from the high real threshold value to the actual threshold value below, are recorded. The pixels that change from the low real threshold value to the high real threshold value are designated as white, while the pixels that change from the actual threshold value above to the low real threshold value are designated as black. Thus, the pairs of pixel addresses are stored for each scan line with the first address of each pair representing the beginning of a segment of white pixels, and the second direction of each pair representing the end of the segment of the white pixels.

After the above, the CPU examines the data encoded in the length path to locate a reference feature of the captured video image. In the embodiment herein, the reference feature is the top flange 102 of the bottle 4 and appears as a bull of contiguous white pixels. Specifically, the CPU performs the Blob detection (large binary object) on the data encoded in the length path to locate the white pixel segments, which resemble the bull. Figure 6 is a negative of the pixels in the window 108 forming the bull (generally indicated with the reference number 1 1 0). Once the bull 10 10 of white pixels is located, a closing rectangle 1 12 is defined aligned with the scanning lines that encompass the bull 1 10. The center of the rectangle 1 12 is calculated and is used as the location of the bull. reference Xref, Yref. After the reference location Xref, Yref, a specific rectangular, central region of interest 120 has been established within the window 108, which encompasses a portion of the threads 104 of the bottle 14, it is determined by moving downwards, from the reference location in the Y direction through a predetermined number of pixels. Once the specific central region of interest 120 is determined, another 10 specific rectangular regions of interest 122 are determined on both sides of the central specific central region of interest 120, giving a total of 21 specific regions of interest. The specific regions of interest 122 are greatly deviated in the Y direction away from the specific central region of interest 120, so that they follow the bottle threads around its circumference. The dimensions of the specific regions of interest 122 are also reduced in the X direction from the central specific region of interest 120 to compensate for the perspective effects. Thus, the numbers of pixels within the specific regions of interest 120 and 122 vary, although each specific region of interest has approximately 1000 pixels (50 x 20 pixels). After the above, the CPU calculates a black / white threshold value for each of the specific regions of interest 120 and 122. The black / white threshold value calculated for each of the specific regions of interest is based on the actual threshold value previously determined. Nevertheless, the actual threshold value is classified for each of the specific regions of interest subtracting an empirical deviation. The empirical deviation is increased for the specific regions of interest 122 plus from the specific central region of interest 120. Thus, the black / white threshold values are reduced by more magnitude in the specific region of interest than from the specific region. central interest. With the threshold values set for the specific regions of interest, the CPU compares the pixels in the specific regions of interest within the window 108 with the threshold value and designates the pixels as white or black depending on the results of the comparisons for generate pixels in binary. The Blob detection is then performed on the binary pixels in each of the specific regions of interest to locate contiguous white pixel groups in each specific region of interest. The CPU then collects a histogram for each specific region of interest by counting the number of pixels in several groups. The areas of contiguous pixel groups detected in the specific regions of interest 122 are then classified according to their positions relative to the central specific region of interest 120 to correct perspective effects, which make a given defect look smaller in the specific external regions of interest 122 that what might appear in the specific central region of interest 120. Groups of pixels having five or fewer contiguous binary pixels are discarded. The binary pixels of the remaining groups are then added together and the sum is compared to a threshold area value. If the sum exceeds the threshold area value, the contiguous white pixel groups are filtered to reduce false detection of the thread defects. In particular, the shape and characteristics of the area of adjacent white pixels are examined to determine if they resemble the threads of a bottle. If a comparison is determined, the CPU determines that the generated white pixels are not the result of a thread defect, but rather the result of a well-defined thread. However, if a comparison is not determined, it is considered that a defect 126 is located on the threads of the bottle (see Figures 7a and 7b). During the above procedure, the bull 10 and the profiles of the specific areas of interest are displayed on the operator monitor 48 superimposed on the video image. Groups of contiguous white pixels are also displayed on the operator monitor 48 within the appropriate profile of the specific area of interest. Once a thread defect is detected, the CPU outputs the captured image to the live monitor 52 via the junction board 30, so that the defective bottle is unfolded. The CPU also sends signals to the production line board 44, so that the defective bottle can be traced. When the defective bottle arrives at the bottle detection sensor 28 and is detected, the production line board 44 sends signals to the bottle reject mechanism 18 via the junction board 30, so that the bottle reject mechanism can track the defective bottle and remove it from the conveyor system 12 when the defective bottle arrives at the bottle rejection mechanism. The procedure described above is performed for the video images captured by each of the video image processors 42. Thus, as each bottle 14 travels along the inspection system 16, four video images of the flange are taken. and the threads of the bottle. Since these video images span approximately 1 10 degrees from the circumference of the bottle and are taken around the circumference of the bottle to spaces of approximately 90 degrees, the entire threaded section of each bottle is examined for defects.

In addition to the above, the PCOM card 46 on the computer 40 checks the output of the temperature sensor 32 through the junction board 30. If the detected temperature exceeds a threshold value, the PCOM card 40 activates the guide 38 through the junction board 30. The PCOM card 46 also checks the junction board 30 to determine if the halogen lamps 24 are energized. If it is determined that any of the halogen lamps 24 is not ejecting current, the PCOM card 46 activates the guide 38 via the junction board 30. Since the inspection system 16 typically operates only, the activation of the guide 38 allows that the conditions of failure in the operation of the inspection system 16 are detected by an operator. Although a particular example of Blob detection has been described to detect white pixels, those skilled in the art will appreciate that other methods may be employed, such as the method described in the G article. H irzinger and K. Landzattel, entitled "A Fast Technique for Segmentation and Recognition of Binary Patterns", IEEE Conference on Pattern Recognition and Image Proccessing, 1981, Run-Lengyh Connectivity Analysis, as described in the article by I. Kabir, entitled "A Computer Vision System U sing Fast One Pass Algorithms", M. S. Thesis, University of California, Davis, 1983, the Clustering Method as described in an article by R. C. Smith and A. Rosenfield, entitled "Threshold Using Relaxation", I EEE Trans. On Pattern Analysis and Machine Intelligence, Vol. 3, p. 598-605, 1981, the Region Development Method (Growing Method Region), as described in an article by C. R. Brice and C. L. Fenneman, entitled "Scene Analysis Using Regions", Artificial Intelligence, Vol. 1, p. 205-226, 1970, and the Split and Merge Method as described in an article by D. M. Mark and D. J. Abel entitled "Linaer Quadtrees from Vector Representation of Plygons", IEE Trans. On Pattern Analysis and Machine Intelligence, Vol. PAMI-7, No. 3, p. 344-349, 1985. As shown, the present invention provides a method of computerized video analysis for detecting defects of threads in glass bottles without the bottles being handled or the speed of the conveyor system 12 being reduced. In addition, using computerized video analysis, the bottles do not need to be placed precisely before the analysis. Since the bottles do not have to be handled, the damage and contamination of the bottles is reduced. Although Figure 6 illustrates 21 collateral regions of interest, each of which is individually processed to detect thread defects, those skilled in the art will appreciate that the number and orientation of specific regions of interest within the video image may to vary. For example, Figures 8 and 9 show an alternative arrangement of specific regions within the video image. In this embodiment, the CPU binaries the pixels within the window to locate the reference feature ((ie, the bottle rim 210) in the same manner as previously described.A lock rectangle 212 is defined aligned with the scanning lines, which encompasses the bottle rim 210, the upper left corner of the closing rectangle is used as the reference location Xref, Yref A specific central region of interest 220 is then determined. The specific central region of interest has been determined, nine specific rectangular regions of interest 222 are determined on both sides of the specific central regions of interest, giving a total of 19 specific regions of interest.The specific regions of interest slightly overlap with the specific adjacent regions. of interest and, as in the previous modality, they deviate greatly in the Y direction from the specific region a central interest to follow the threads of a bottle in the video image. Once the specific regions of interest 220, 222 have been determined, the CPU calculates a threshold value for each of the specific regions of interest. With the threshold values set for the specific regions of interest, the C PU compares the pixels in the specific regions of interest with the threshold values to create pixels with binaries in a manner similar to that previously described. Following this and unlike the previous modality, the C PU groups the 19 specific regions of interest into five groups 225, depending on the relative positions of the specific regions of interest, with respect to the central region of interest. The Blob detection is then performed on groups 225 of pixels with binaries to locate contiguous white pixels in each group 225. The CPU then collects a histogram for each group 225 by counting the number of contiguous white pixels. The histograms collected for groups 225 afterwards are classified according to their positions in relation to the specific central region of interest to correct the effects in perspective. Once classified, the number of contiguous white pixels in the groups is compared with a threshold area value. If the contiguous white pixel numbers are greater than the threshold area value, the contiguous white pixels are filtered in the previously described manner before adjacent white pixels are determined to represent the thread defects. Referring now to Figures 10 to 13, an alternative embodiment of a bottle production line 310 is shown, including an inspection system 316 for detecting defects in bottles. As can be seen, the inspection system 316 includes a read star 400, which receives a separate bottle stream 314 from a feed screw 402 at the outlet end of an input conveyor 404. The read star 400 delivers the bottles to the cavities of the main reading star 406, which incorporates backup fasteners 408 to secure the bottles to the main reading star and prevent the bottles from rotating or vibrating. The main reading star 406 in turn supplies the bottles to the cavities of another reading star 410, which delivers the bottles either to an output conveyor 412 or to a bottle reject conveyor 414. The image forming sections video 320 are arranged around the periphery of the main reading star 408 at circumferentially spaced locations to inspect a different portion of each bottle as it moves through the main reading star. Each video image section includes a photoelectric bottle sensing sensor 326, which detects the arrival in a bottle. The bottle detection sensor 326, when activated, sends signals to a computer 340, which in turn activates a strobe light 324. The CCD camera 322 is associated with each strobe to take a picture of the bottle illuminated against light, when the strobe light is activated. The video image processors 342 capture and process the video images taken by the CCD cameras 332, when the strobe lights are operated. The computer 340 is connected to an operator monitor 348 and a central control system 426, so that the operation of the inspection system can be co-oriented with the operation of the rest of the bottle installation. The computer 340 also communicates with a programmable logic controller (PLC) 428. The PLC 428 communicates with activation controllers 430, associated with the read stars 400, 406 and 410, backup fasteners 408, input advance screw 402 and staples 41 1 of the reading star. Referring now to Figure 12, the orientation of the CCD camera 322 and the strobe light 324 in one of the video image forming sections is further illustrated. As can be seen, the CCD camera 322 is oriented, so that its optical axis points down to the bottle at an angle equal to 45 ° with respect to the horizontal H. The strobe light 324 is similarly oriented, so that it directs the light towards the bottle at an angle inclined with respect to the horizontal H equal to approximately 30 °. The captured video images are processed by the video image processors 342 in a manner similar to that previously described, to detect bottleneck defects, as illustrated in Figure 13. As can be seen, in this embodiment, the CPU determines five specific collateral regions of interest 440 following the threads of the bottles in the video image. In this embodiment, when a defective bottle is detected by the CPU within the computer 340, the computer 340 controls the read star 410, so that a bottle is dispensed to the defective bottle conveyor 414 instead of the output conveyor. 412. In this way, defective bottles are removed from bottle production line 310. Although inspection systems have been described as including four video image forming sections, those skilled in the art will appreciate that few can be used. video or additional image forming sections, depending on the desired accuracy. Also, although each video image processor has been described as having a dedicated CPU for processing the captured video images, it should be understood that the number of mother cards can be reduced with each video image processed by the motherboard, captured by more than one secondary card. Also, although a bottle detection sensor is shown for each video image forming section, it should be understood that an individual bottle detection sensor can be used to detect each bottle as it approaches the inspection system. In this case, the output of the encoder is used to determine the speed of the bottles, and the activation of the CCD cameras is based on the determined speed of the bottles and the separation between successive video image forming sections. Although specific embodiments of the present invention have been described, those skilled in the art will appreciate that variations and / or modifications can be made without departing from the scope thereof, as defined in the appended claims.

Claims (4)

1. CLAIMS 1 .- A method for inspecting a bottle having a threaded section for thread defects as said bottle moves along a production line, comprising the steps of: (i) capturing a video image of said bottle with a video camera as the bottle moves towards the field of vision of said video camera without requiring that said bottle be in a specific position within the field of vision, said video image encompassing a general region of interest that it contains at least a portion of the threaded section of said bottle; (ii) determining the position of the bottle within the general region of interest based on the location of a feature of said bottle within the general region of interest; (iii) segmenting a portion of said general region of interest, to which it encompasses at least a portion of the threaded section in a plurality of specific regions of interest; and (iv) examining the pixels of said video image in the specific regions of interest to detect thread defects.
2. 2. The method according to claim 1, wherein said feature is the upper flange of said bottle.
3. 3. The method according to claim 2, further comprising the steps of establishing a reference location based on the location of said top flange; and detecting the portion of said general region of interest and the positions of said regions of interest based on the position of the reference location.
4. 4. The method according to claim 3, wherein the position of the central specific region of interest is determined relative to said reference location, and wherein the positions of other specific regions of interest on opposite sides of the region. specific central interest are determined in relation to the position of the specific central region of interest. 5. - The method according to claim 4, wherein said other specific regions of interest are greatly deviated in a Y direction from the specific central region of interest to follow the threaded section of the bottle in the video image. 6 - The method according to claim 5, wherein the specific regions of interest are reduced in an X direction from the central specific region of interest to compensate for the perspective effects. 7 - The method according to claim 6, wherein the specific regions of interest are collaterally positioned. 8. - The method according to claim 6, wherein the specific regions of interest overlap adjacent specific regions of interest. 9. The method according to claim 3, wherein the step of examining said pixels includes the steps of: determining a black / white pixel threshold value for the specific regions of interest; comparing said pixels with the threshold value and biasing said pixels as white or black, depending on the results of each comparison; and determining contiguous white pixel groups greater than a threshold value, to detect thread defects. 10. The method according to claim 9, wherein during the step of determining contiguous white pixel groups, adjacent white pixel groups are filtered to reduce false detection of bottle thread defects. 1 - The method according to claim 1, wherein steps (i) to (iv) are performed in a plurality of circumferentially spaced locations around said bottle to detect thread defects throughout the threaded section of the bottle . 12. A system for inspecting bottles having a threaded section as the bottles move along a production line, comprising: a plurality of video image forming sections, arranged along the line of production in separate locations, each video image forming section being oriented with respect to said production line to take a video image of each bottle in a different circumferential region thereof, as each bottle moves into the field of view of the video image forming section without requiring the bottle to be in a specific location in the field of view, said video images encompassing a general region of interest containing by at least a portion of the threaded section of each bottle; and processing means in communication with the video image forming sections and receiving the video images taken by them, said processing means processing each video image to determine the position of the bottle within the general region of interest based on the location of a feature of said bottle within the general region of interest; segmenting a portion of the general region of interest, which encompasses at least a portion of the threaded section to a plurality of specific regions of interest; and examining the pixels in the specific regions of interest to detect thread defects. 13. A system for inspecting bottles according to claim 12, wherein each video image forming section includes a video image forming chamber and a light source, the light source and the video image forming chamber. are located on opposite sides of the production line, the video image forming chambers and the light sources in said video image forming sections are oriented so that the video images of the entire circumference of each bottle are taken by the training sections of video. 14. - A system for inspecting bottles according to claim 13, further comprising at least one bottle detection sensor for detecting the presence of a bottle, said processor means responding to at least one bottle detection sensor and activating said bottles. video image forming sections to take video images of said bottle. 15. A system for inspecting bottles according to claim 14, wherein said light sources remain illuminated, and wherein said activating means activates the video image forming chamber to take an image of said bottle in response to the sensor of said image. associated bottle detection. 16. A system for inspecting bottles according to claim 15, wherein said processing means sends signals to a bottle rejection mechanism along said production line after detecting a bottle with a thread defect. 17. A system for inspecting bottles according to claim 13, wherein said video image forming sections are arranged in closing pairs to reduce the separation and the interference between them. 18. A system for inspecting bottles having a threaded section as the bottles move along a production line, comprising: a plurality of video image forming sections disposed along said production line in separate locations, each video image forming section being oriented with respect to said production line to take a video image of each bottle in a different circumferential region thereof, as each bottle moves towards the field of view of the video image forming section without requiring the bottles to be in a specific location in the field of view, said video image forming sections being arranged to reduce the spacing between them; and processing means in communication with the video image forming sections and receiving the video images taken by them, said processing means processing the video images to detect thread defects in said bottle. 19. - A system for inspecting bottles according to claim 18, wherein each video image forming section includes a video image forming chamber and a light source, the video image forming chamber and the light source being positioned on opposite sides of the production line and being laterally offset, so that the optical axis of each video image forming section forms an oblique angle with respect to said production line. 20. - A system for inspecting bottles according to claim 19, wherein said system includes four video image forming sections arranged in an upstream pair and a downstream pair, the video image forming chambers and the sources of light in said upstream pair forming an obtuse angle with respect to the production line and the video image forming chambers and the light sources in the downstream pair forming acute angles with respect to said production line. 21. A system for inspecting bottles according to claim 20, wherein the positions of the video image forming chambers and the light sources with respect to the production line alternate in successive video image forming sections. 22. A method for locating the position of a bottle within a video image, comprising the steps of: capturing a video image of at least a portion of said bottle with a video camera; digitize said video image; compare the pixels in the video image with a threshold value and put in binaries said pixels as white or black depending on the results of said comparisons; and determining the position of said bottle relative to the boundaries of the video image based on the location of the largest group of contiguous white pixels within said video image. 23 - A method for detecting defects in bottle threads, comprising the steps of: capturing a video image of said bottle threads; segmenting at least a portion of the video image to a plurality of specific regions of interest; and process the video of each specific region of interest, independently, to determine defects in each of the specific regions of interest, based on the existence of white areas greater than a prescribed area size threshold within said specific regions of interest . 24. A system for locating the position of a bottle within a video image, comprising: means for illuminating said bottle with a light source against light; a video camera for capturing a video image of at least a portion of the bottle within the field of view thereof; means for digitizing said video image; and means for determining the position of the bottle within the video image, based on the location of the largest white area within said video image. 25 - A system for detecting defects in bottle threads, comprising: means for capturing a video image of the bottle threads; means for segmenting at least a portion of the video image to a plurality of specific regions of interest; and video processing means of each specific region of interest, independently, to determine the defects in each of the specific regions of interest based on the existence of white areas greater than a prescribed area size threshold within the specific region of interest. interest.