JP2007133715A - Method and apparatus for processing image - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a method and an apparatus for processing images, enabling improvement of inspection efficiency, as well as reduction of a processing time and miniaturization, even a certain amount of deviation between a product inspection range and an imaging range is produced at the time of imaging. <P>SOLUTION: Even if a certain amount of deviation is produced between the inspection range of a product P and the imaging range at the time of imaging, each of n consecutive images imaging the product P is stored in a (1/n) area, an area for one image, of an image memory 3, and decision is made thereupon. Accordingly, the processing time can be reduced because readout of only the image in the (1/n) area is sufficient in each, and miniaturization can be attained because the n images are stored in the area for one image of the image memory 3. Furthermore, through the decision based on n images deviating before and after at the time of imaging, tolerance between the inspection range and the imaging range can be increased. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention relates to an image processing method and apparatus for performing image processing on an imaged product to determine the quality of the product, and more particularly to improving inspection efficiency when a slight shift occurs between the inspection range of the product and the imaging range during imaging.

  As shown in FIG. 1, as a conventional image processing apparatus, imaging data of a product P imaged by an imaging camera (image sensor) 2 while conveying a number of products P one after another by a conveyance mechanism 4 such as a conveyance conveyor. An image processing unit 5 that performs image processing to determine the quality of a product P is known. Usually, since a predetermined time (for example, several tens of milliseconds) is required for image processing of the imaged product P, in order to give an imaging timing to the imaging camera 2, a photoelectric that detects the passage of each product P and outputs an imaging request signal A synchronous sensor 1 such as a sensor is used.

  As shown in FIG. 6A, when the imaging camera (image sensor) 2 captures an imaging target part of the product P, the inspection range and the imaging range of the imaging camera 2 are installed so as to coincide with each other. When an imaging request signal from the sensor 1 is input, imaging is performed at an imaging timing based on this input.

  By the way, as shown in FIG. 6B, when the posture of the product P is slightly inclined due to the vibration of the transport conveyor 4 or the like during imaging, the input timing of the imaging request signal from the synchronization sensor 1 is shifted and the imaging timing is changed. Since the image pickup range and the inspection range do not match, the non-defective product is judged as a defective product, and the inspection efficiency is lowered.

  In such a case, conventionally, there has been known a method of widening the imaging range of the imaging camera 2 using a high-resolution image processing apparatus so as to allow a slight shift between the imaging range and the inspection range. There is a problem of high cost of the apparatus. In addition, there is known a method of expanding an imaging range by changing an optical system of a camera that captures an image. However, since an inspection object is captured in a small size, accuracy is lowered and a false detection rate is increased.

On the other hand, it is also known that for each shift of a plurality of products, imaging is performed a plurality of times after the synchronization sensor is activated, and the image data is statistically processed to calculate an average value or the like (for example, patent document) 1). A method is also known in which an image is captured a plurality of times by the operation of a synchronous sensor, the image is stored in a large-capacity memory, and processed later (for example, Patent Document 2).
JP 2004-145504 A JP-A-8-70420

  However, since Patent Document 1 performs statistical processing of these image data after performing image processing that requires a predetermined time for a plurality of products a plurality of times, the processing takes time. Further, Patent Document 2 requires a large-capacity memory for storing a plurality of images, which increases the size.

  The present invention solves the above-described problems, and can improve the inspection efficiency while reducing the processing time and reducing the size even if there is a slight deviation between the inspection range of the product and the imaging range during imaging. An object of the present invention is to provide an image processing method and apparatus.

  In order to achieve the above object, an image processing method and apparatus according to the present invention determines an inspection target part of a product based on an input of an imaging request signal for requesting imaging for each of a plurality of products sequentially transported by a transport mechanism. The inspection range and the imaging range are matched to capture an image while scanning with an image sensor, and the captured image data of the region to be inspected is image-processed and stored in an image memory, and the quality of the product is determined based on this image. In response to one input of an imaging request signal, n (n is an integer greater than or equal to 2) images are continuously captured, and each image is (1 / n) of the area for one image in the image memory. By storing each image in the area, n images are stored in the area corresponding to one image, and the product OK / NG is determined for each of the stored n images. Determining a product OK when the one image of OK or Re, determines a product NG, the first process of performing in the case of NG for all images.

  According to this configuration, even if there is a slight deviation between the inspection range and the imaging range of the product P during imaging, the (1 / n) area of one image in the image memory for consecutive n images of the product. Therefore, it is only necessary to read out each (1 / n) area image, so that the processing time can be shortened and n images are stored in an area of one image in the image memory. Since it is stored, the image memory can be reduced in size. In addition, it is possible to increase the tolerance of the shift between the inspection range and the imaging range by the determination based on n images before and after the shift at the time of shooting. Therefore, even if there is a slight deviation between the product inspection range and the imaging range during imaging, the inspection efficiency can be improved while shortening the processing time and reducing the size of the apparatus.

  Preferably, in the first process, two images are continuously captured in response to the input of the one imaging request signal, and an even-numbered scan row is set for one image and an odd-numbered scan row is set for the other image. These two images are respectively thinned and stored in (1/2) area of one image memory area, and the product OK / NG is determined for each of the two stored images. Therefore, the processing time can be shortened and the apparatus can be downsized with a simple structure.

  Preferably, in response to the first processing and one imaging request input, one image is captured, this image is stored in an area of one image in the image memory, and the stored one image It is possible to switch between the second process for determining whether the product is OK / NG. Therefore, it is possible to cope with a case where there is no deviation between the inspection range and the imaging range as necessary.

Hereinafter, embodiments of the present invention will be described with reference to the drawings.
FIG. 2 is a block diagram showing an image processing apparatus according to an embodiment of the present invention. This apparatus detects, for a plurality of products P sequentially transported by a transport mechanism 4 (for example, the transport conveyor of FIG. 1), outputs each image P and requests an image capturing request signal, such as a synchronous sensor. An imaging camera (image sensor) that captures an image of an inspection target part of product P while scanning the inspection range and the imaging range in accordance with the imaging request signal output unit 1 and the input of the imaging request signal from the synchronization sensor 1 2 and an image processing unit 5 that performs image processing on captured image data of the product P, stores the image in the image memory 3, and determines the quality of the product P based on the image. The positions of the synchronization sensor 1 and the imaging camera 2 are fixed. For example, a CCD camera is used for the imaging camera 2, and a reflective photoelectric sensor is used for the synchronization sensor 1.

  As shown in FIG. 6 (A), for example, the product P is food stored in a packaging box, and the inspection target part of the product P has a letter of expiration date (two-stage date) of the food on the packaging box. The printed character is the inspection range. The imaging camera 2 is installed so that the inspection range and the imaging range coincide with each other, and images this. In the present invention, when the imaging range does not coincide with the inspection range due to a difference in imaging timing due to a slight inclination of the product P during imaging (FIG. 6B), the imaging range and the inspection range are slightly different. The inspection efficiency is improved by allowing the deviation and reducing the erroneous inspection of a non-defective product as a defective product as much as possible.

  As shown in FIG. 2, the image processing unit 5 includes the image memory 3, the storage processing unit 6, the determination unit 7, and the memory 9. The storage processing unit 6 continuously captures n (n is an integer of 2 or more) images in response to one input of the imaging request signal, and each image is stored in an area of one image in the image memory 3 ( 1 / n), each image is stored, and all images are stored in the area for one image. The memory 9 stores character information, a reference image, and the like in the inspection target part of the product P.

  The determination unit 7 reads out each of the images stored in the (1 / n) area of the n images stored in the area for one image of the image memory 3, and reads the characters based on the binary information. Each product OK / NG is determined by a known image processing method such as recognition or pattern matching with a reference image stored in the memory 9. As a result, the determination unit 9 determines that the product is OK when at least one of the images is OK, and determines that the product is NG when all the images are NG. The first processing is performed by the operations of the storage processing unit 6 and the determination unit 7.

  In this example, as the first processing, the storage processing unit 6 continuously captures two images in response to one input of the imaging request signal, scans even rows for one image, and the other image Each of the odd-numbered scans is thinned out, and these two images are stored in each (1/2) area of the area of one image in the image memory. Two images, which are thinned out every other row and whose vertical resolution is halved, are read out while being stored in the (1/2) area, and the product OK / NG is determined.

  In addition, the image processing unit 5 captures one image in response to the first processing and one input of the imaging request signal, and stores this image in an area for one image of the image memory 3, A process switching unit 8 capable of switching between the second process for determining the product OK / NG for one stored image is provided. The process switching unit 8 is configured to prevent the posture of the product P from being tilted, for example, when the posture does not tilt depending on the type of the product P, or by providing a guide mechanism for guiding the product P on the transport conveyor 4. Thus, when the imaging timing does not shift, the process is switched to the second process.

  An example of the image processing operation in the apparatus having the above configuration will be described based on the flowchart of FIG. First, when the product P passes, the synchronous sensor 1 is activated and detected, and an imaging request signal is output to the image processing unit 5 (step S1). Based on the input of the imaging request signal, the imaging camera 2 starts the first imaging (step S2). After the completion of the first imaging, the first image is stored only in an even numbered line (1/2 area) in an area of one image in a predetermined area of the image memory 3, that is, a thinning operation is performed to discard the odd numbered lines. (Step S3).

  Next, the imaging camera 2 starts the second imaging continuously (for example, at an interval of 13 msec) after the first imaging (step S4). After completion of the second imaging, the second image is stored only in the odd-numbered rows (1/2 area) in the area of one image in the area of the image memory 3, that is, a thinning operation is performed to discard the even-numbered rows. (Step S5). FIG. 4 shows a storage state of, for example, “2” which is a part of the character of the inspection target part of the product P in the area of one image in the image memory 3. A state in which the image is stored in the even-numbered rows and (B) shows a state in which the second image is stored in the odd-numbered rows. In both images, the odd-numbered or even-numbered scans are thinned out and the vertical resolution is halved. In this figure, the first image and the second image are drawn in the same way, but in reality, there is a time difference (13 msec) between the imaging timings of both images, so depending on the transport speed of the transport conveyor 4 Thus, the image of (B) is slightly shifted from the image of (A). The time difference between the imaging timings of the two images is set in accordance with the timing of occurrence of the deviation of the posture of the product P during imaging according to the speed of the conveyor 4. As a result, the two images before and after the shift are stored in the area of one image in the image memory 3.

  Next, image processing such as character recognition and pattern matching is performed on the first even-numbered line image in a state where the vertical resolution is ½, and then read out from the image memory 3 (step). S6) The quality of the product P is determined for the first image (step S7). If the first image is OK, the product OK is output (step S10). In the case where the first image is NG in step S7, the second odd-numbered row image is similarly processed and read out (step S8), and the quality of the product P is determined for the second image. Is determined (step S9). If the second image is OK, the product OK is output (step S10). In the case of NG for the second image, since all the two images are NG, the product NG is output (step S11). Thus, the first process is performed.

  If the posture of the product P or the product P is not tilted and the posture is not tilted and the imaging timing is not shifted, the process switching unit 8 is switched as shown in the flowchart of FIG. The following second process is performed.

  First, when the product P passes, the synchronous sensor 1 is activated and detected, and an imaging request signal is output to the image processing unit 5 (step S31). Based on the input of the imaging request signal, the imaging camera 2 starts imaging (step S32). After completion of imaging, the image is stored in an area for one image in a predetermined area of the image memory 3 (step S33).

  Next, the image is processed in the same manner and then read out from the image memory 3 (step S34), and the quality of the product P is determined (step S35). If this image is OK, the product OK is output (step S36). If the image is NG, the product NG is output (step S37). Thus, the second process is performed.

  In this way, the processing switching unit 8 can be switched to cope with a case where there is no shift in the imaging timing because there is no inclination of the posture of the product P, and there is no shift in the imaging range with respect to the inspection range. In this case, since the second process is stored and determined for only one image as compared with the first process, the processing time can be further shortened.

  As described above, according to the present invention, even if there is a slight deviation between the inspection range and the imaging range of the product P during imaging, two images obtained by imaging the product P are equivalent to one image in the image memory 3. Since each image is stored and determined in each (1/2) area, it is only necessary to read out the image in the (1/2) area. Therefore, the processing time can be shortened and two images are stored in the image memory 3. Therefore, the image memory 3 can be reduced in size. In addition, although the odd-numbered or even-numbered scans are thinned out and the vertical resolution is halved, the determination based on the two images before and after the shift at the time of shooting allows the tolerance of the shift between the inspection range and the imaging range. Can be increased. Therefore, even if a slight deviation occurs between the inspection range of the product P and the imaging range during imaging, the inspection efficiency can be improved while shortening the processing time and reducing the size of the apparatus.

  In the above embodiment, when two images are stored in the area of one image of the image memory 3, the scan even rows are thinned out for either one image, and the scan odd rows are thinned out for the other image. However, the present invention is not limited to this. For example, one of the images may be compressed and stored in the upper half area for one image, and the other image may be stored in the lower half area. Also good. However, compression processing takes more processing time than decimation.

  In the above-described embodiment, two continuous images are taken, two images are stored in an area for one image, and the determination is made for two images. For example, three or more images may be stored in an area for one image, and three or more images may be determined.

  In the above embodiment, the process switching unit 8 for switching between the first process and the second process is provided. However, the process switching unit 8 may be omitted by performing only the first process.

It is a block diagram of the image processing apparatus provided with the conveyance conveyor, the synchronous sensor, the imaging camera, and the image process part. 1 is a block diagram illustrating an image processing apparatus according to an embodiment of the present invention. It is a flowchart figure which shows the image processing operation by a 1st process. It is a block diagram which shows the state in which each image was stored in the even-numbered line or the odd-numbered line of the image memory. It is a flowchart figure which shows the image processing operation by a 2nd process. It is a block diagram which shows the agreement and discrepancy of an image range and an imaging range.

Explanation of symbols

1: Imaging request signal output unit (synchronous sensor)
2: Image sensor (imaging camera)
3: Image memory 4: Conveyance mechanism (conveyor)
5: Image processing unit 6: Storage processing unit 7: Determination unit 8: Process switching unit P: Product

Claims (6)

Based on the input of an imaging request signal for requesting imaging for each of a plurality of products sequentially transported by the transport mechanism, the inspection target part of the product is imaged while being scanned with an image sensor so that the inspection range matches the imaging range, An image processing method for image-capturing imaging data of a region to be inspected and storing it in an image memory, and determining the quality of a product based on the image,
In response to a single input of an imaging request signal, n (n is an integer of 2 or more) images are taken continuously, and each image is respectively (1 / n) of one image memory area. By storing, n images are stored in an area corresponding to the one image, and the product OK / NG is determined for each of the stored n images, and as a result, at least one of the images is determined. An image processing method for performing a first process of determining that the product is OK in the case of OK and determining that the product is NG in the case of NG for all the images. In claim 1,
In the first process, two images are continuously captured in response to the input of the one imaging request signal, and the even-numbered scan row is thinned out for one image and the odd-numbered scan row is thinned out for the other image. Then, these two images are respectively stored in the (1/2) area of the area of one image in the image memory, and the product OK / NG is determined for each of the two stored images. . The claim 1, further comprising:
The first process;
In response to a single imaging request input, one image is captured, this image is stored in an area for one image in the image memory, and the second OK / NG determination of the product is performed for the stored one image. An image processing method that makes it possible to switch between these processes. An imaging request signal output unit that outputs an imaging request signal for requesting imaging for each of a plurality of products that are sequentially conveyed by a conveyance mechanism, and an inspection range and an imaging range of an inspection target portion of the product based on the input of the imaging request signal An image sensor including an image sensor that scans images while matching them, and an image processing unit that performs image processing on the captured image data of the examination site and stores the image data in an image memory, and determines the quality of the product based on the image A processing device comprising:
The image processing unit includes a storage processing unit and a determination unit,
The storage processing unit continuously captures n (n is an integer of 2 or more) in response to a single imaging request signal input, and each image is (1 / of the area of one image in the image memory. n) By storing each in the area, n images are stored in the area for one image,
The determination unit determines the product OK / NG for each of the stored n images, and as a result, determines that the product is OK when at least one of the images is OK, and the product when all the images are NG. An image processing apparatus that performs a first process of determining NG. In claim 4,
The first process includes
The storage processing unit continuously captures two images in response to the one imaging request input, thins out even-numbered scans for one image, and scans odd-numbered rows for the other image. Two images are stored in the (1/2) area of one image area of the image memory,
An image processing apparatus in which the determination unit determines a product OK / NG for each of the two stored images. In claim 4, further:
The first process;
In response to one imaging request input, one image is captured, this image is stored in an area of one image in the image memory, and the product OK / NG is determined for this stored one image. And processing
An image processing apparatus provided with a process switching unit capable of switching between.

Images