JP2020148586A - Image processing apparatus, inspection device, image processing method, and program - Google Patents

Abstract

To make it possible, in appearance inspection through image processing, to accurately and simply extract an area where an inspection object is captured.SOLUTION: There is provided an inspection device 1 comprising: a recognition unit 110 that, in order to extract an extraction object area corresponding to an inspection object from an input image, recognizes a first partial area inside the extraction object area and a first frame outside the extraction object area for surrounding the extraction object area; a contour candidate specification unit 120 that, based on the first partial area and the first frame, specifies a contour candidate for the extraction object area so as to include the first partial area and not to include the first frame; an update unit 140 that, based on the contour candidate, creates a second partial area inside the contour candidate and a second frame outside the contour candidate for surrounding the contour candidate; an area specification unit 150 that, based on the second partial area and the second frame, specifies the extraction object area so as to include the second partial area and not to include the second frame; and an appearance inspection unit 160 that performs appearance inspection with reference to the specified extraction object area in the input image.SELECTED DRAWING: Figure 2

Description

The present invention relates to an image processing apparatus, an inspection apparatus, an image processing method, and a program.

The visual inspection is an inspection for determining the quality of an article by confirming defects such as foreign matter, dirt, scratches, burrs, chips, and deformation adhering to the surface of an article mass-produced industrially. Conventionally, the visual inspection is generally performed visually by a person, but in recent years, it has been proposed to realize the visual inspection by image processing the captured image of the inspection object. Specifically, the appearance inspection is realized by designating the inspection target area in which the inspection target is reflected in the captured image of the inspection target and performing image processing such as shape recognition on the inspection target area. To. It is preferable that the inspection area is specified easily. As an example of a technique for designating a specific area for an image and performing image processing on the area, the technique disclosed in Patent Document 1 can be mentioned. Patent Document 1 discloses a method of easily obtaining a subject area by roughly instructing a region of a subject in an image and its vicinity and performing subject recognition processing.

Japanese Unexamined Patent Publication No. 7-322504

When visual inspection of mass-produced articles is performed by image processing, it is not realistic to specify the inspection target area for each of the captured images of the articles, and each imaging is performed according to one preset rough instruction. It is common to extract the inspection target area from the image. However, the appearance of the image around the inspection target is not always the same for each captured image, and extracting the inspection target region according to a rough instruction may induce erroneous recognition of defects.

In order to solve the above problems, the image processing apparatus according to the present invention is an image processing apparatus that specifies an extraction target region for extracting an image from an input image, and is a first partial region within the extraction target region. A recognition unit that recognizes a first frame that is outside the extraction target area and surrounds the extraction target area, and the first partial region and the first partial region based on the first partial region and the first frame. A contour candidate specifying unit that specifies a contour candidate of the extraction target area so as not to include a frame, a second partial region within the contour candidate based on the contour candidate, and the contour candidate outside the contour candidate. Based on the update unit that generates the second frame surrounding the second frame, the second partial area, and the second frame, the extraction target area is specified so as to include the second partial area and not include the second frame. It is characterized by having a region specifying portion.

Further, in order to solve the above problems, the inspection apparatus according to the present invention includes a first partial region in the extraction target region and the extraction target region in order to extract an extraction target region corresponding to the inspection target in the input image. A recognition unit that recognizes a first frame that is outside the extraction target area and surrounds the extraction target area, and the first frame that includes the first partial area based on the first partial area and the first frame. A contour candidate specifying unit that specifies a contour candidate of the extraction target region so as not to include the contour candidate, a second partial region within the contour candidate based on the contour candidate, and a contour candidate that is outside the contour candidate and surrounds the contour candidate. An area specification that specifies the extraction target area so as to include the second partial area and not include the second frame based on the update unit that generates the second frame, the second partial area, and the second frame. It is characterized by including a unit and an appearance inspection unit that performs an appearance inspection of the inspection object by referring to an image of the extraction target region already specified by the region identification unit in the input image.

Further, in order to solve the above problems, the image processing method according to the present invention is an image processing method for specifying an extraction target area for extracting an image from an input image, and is a first portion in the extraction target area. Recognizing the region and the first frame outside the extraction target area and surrounding the extraction target area, and including the first partial region and the first frame based on the first partial region and the first frame. The contour candidate of the extraction target area is specified so as not to include, and based on the contour candidate, a second partial region within the contour candidate and a second frame outside the contour candidate and surrounding the contour candidate are formed. It is characterized in that the extraction target region is specified so as to include the second partial region and not to include the second frame based on the second partial region and the second frame.

Further, in order to solve the above problems, the program according to the present invention uses a computer as a first partial region in the extraction target region in order to specify an extraction target region of the input image to be extracted. A recognition unit that recognizes a first frame that is outside the extraction target area and surrounds the extraction target area, and the first partial region and the first partial region based on the first partial region and the first frame. A contour candidate specifying unit that specifies a contour candidate of the extraction target area so as not to include a frame, a second partial region within the contour candidate based on the contour candidate, and the contour candidate outside the contour candidate. Based on the update unit that generates the second frame surrounding the second frame, the second partial area, and the second frame, the extraction target area is specified so as to include the second partial area and not include the second frame. It is characterized in that it functions as an area identification unit.

It is a block diagram which shows an example of the hardware structure of the inspection apparatus 1 which concerns on one Embodiment of this invention. It is a functional block diagram for demonstrating the function of the inspection apparatus 1. It is a figure for demonstrating the watershed method. It is a figure for demonstrating the watershed method. It is a figure for demonstrating the watershed method. It is a flowchart which shows the flow of the image processing method executed in inspection apparatus 1. It is a figure which shows an example of the image including the extraction target area. It is a figure which shows an example of the 1st partial area and 1st frame designated by the user of inspection apparatus 1. It is a figure which shows an example of the contour candidate specified by the contour candidate identification processing SA110. It is a figure which shows an example of the 2nd partial area and 2nd frame generated by the update process SA120. It is a figure for demonstrating the effect of this embodiment. It is a figure for demonstrating the effect of this embodiment. It is a figure for demonstrating the effect of this embodiment. It is a figure for demonstrating the modification (3). It is a figure for demonstrating the modification (3). It is a figure for demonstrating the modification (3). It is a figure for demonstrating the modification (3).

Hereinafter, embodiments will be described with reference to the drawings. However, in each figure, the dimensions and scale of each part are appropriately different from the actual ones. Further, although the embodiments described below are subject to various technically preferable limitations, the embodiments are not limited to these embodiments.

A. The first embodiment is a block diagram showing a hardware configuration example of the inspection device 1 according to the embodiment of the present invention. The inspection device 1 is a device for inspecting the appearance of industrially mass-produced articles. As shown in FIG. 1, the inspection device 1 includes a processing device 10, an imaging device 20, a display device 30, an input device 40, a storage device 50, a bus 60 that mediates data transfer between these components, and a bus 60. including.

The processing device 10 is a processor that controls the entire inspection device 1, and is, for example, a computer composed of a single or a plurality of chips. The processing device 10 is composed of, for example, a central processing unit (CPU) including an interface with peripheral devices, an arithmetic unit, a register, and the like. The processing device 10 functions as a control center of the inspection device 1 by executing the control program 510 stored in the storage device 50. The details of the processing executed by the processing apparatus 10 according to the control program 510 will be clarified later.

The image pickup device 20 is, for example, a CCD (Charge Coupled Device) image sensor. The image pickup apparatus 20 is for capturing an image of an inspection object for visual inspection. Hereinafter, the image of the inspection object is referred to as an inspection object image. The image pickup apparatus 20 takes an image of the inspection object and gives the inspection object image data representing the inspection object image to the processing device 10 via the bus 60.

The display device 30 is, for example, a liquid crystal display. The display device 30 displays various images under the control of the processing device 10.

The input device 40 includes a pointing device such as a mouse and a keyboard. The input device 40 is for causing the user of the inspection device 1 to perform various input operations. When the user performs some input operation such as a key operation on the input device 40, the input device 40 gives operation content data representing the operation content of the user to the processing device 10 via the bus 60. As a result, the operation content of the user is transmitted to the processing device 10.

Although detailed illustration is omitted in FIG. 1, the storage device 50 includes a volatile storage unit and a non-volatile storage unit. A volatile storage unit is a storage device that cannot retain stored information unless electric power is supplied, and a non-volatile storage unit retains stored information even if electric power is not supplied. A storage device that can be used. An example of a volatile storage unit is a RAM (Random Access Memory). The volatile storage unit is used by the processing device 10 as a work area when executing the control program 510. A hard disk is an example of a non-volatile storage unit. The control program 510 is stored in advance in the non-volatile storage unit.

The processing device 10 reads the control program 510 from the storage device 50, for example, when the power of the inspection device 1 is turned on (not shown), and starts executing the control program 510. FIG. 2 is a diagram showing an example of a functional block realized by the processing device 10 according to the control program 510. As shown in FIG. 2, the processing device 10 operating according to the control program 510 includes a display control unit 100, a recognition unit 110, a contour candidate identification unit 120, an invalidation unit 130, an update unit 140, and an area identification unit 150. And function as a visual inspection unit 160. The functions of each of these parts are as follows.

The display control unit 100 causes the display device 30 to display various images. Examples of the image displayed on the display device 30 by the display control unit 100 include an image of an inspection object and an image of a contour candidate specified by the contour candidate specifying unit 120 described later. The user who visually recognizes the inspection target image displayed on the display device 30 has a first partial area in the inspection target area in which the inspection target is reflected and a first portion outside the inspection target area that surrounds the inspection target area. The frame is designated by an operation on the input device 40 such as a mouse click. As an example of the mode in which the user designates the first subregion, there is a mode in which the mouse drag is performed so as to draw the outline of the region designated as the first subregion. Similarly, the first frame may be specified by dragging the mouse. When the first partial area is designated, the display control unit 100 causes the display device 30 to display the designated first partial area in a manner that is easy for the user to understand. Further, when the first frame is designated, the display control unit 100 causes the display device 30 to display the designated first frame in a manner that is easy for the user to understand. As the display mode of the first partial area and the first frame element, a mode of displaying in a display color different from that before each designation, a mode of blinking the first partial area and the first frame in different blinking patterns, and the like. Can be mentioned.

The recognition unit 110 recognizes the first subregion designated by the user and the first frame also designated by the user. When the first partial region is designated by dragging the mouse, the recognition unit 110 recognizes the region inside the closed locus represented by the data output from the input device 40 in the process of dragging the mouse as the first partial region. Similarly, when the first frame is designated by dragging the mouse, the recognition unit 110 recognizes the first frame as the closed locus represented by the data output from the input device 40 in the process of dragging the mouse. Although the details will be described later, since the inspection target area is the extraction target by the area identification unit 150, the inspection target area may be referred to as the extraction target area below.

The contour candidate identification unit 120 includes the first partial area and does not include the first frame based on the first partial region recognized by the recognition unit 110 and the first frame recognized by the recognition unit 110. Identify contour candidates for the extraction target area. More specifically, the contour candidate identification unit 120 binarizes the image to be extracted in the extraction target area, and for each row of pixels arranged in the horizontal scanning direction and for each column of pixels arranged in the vertical scanning direction. The contour candidate is specified by performing the processing of the horizontal method.

3 to 5 are diagrams for explaining the watershed method. For example, suppose that the brightness of the pixels belonging to a certain row changes along the row as shown in FIG. For example, it is assumed that the inspection object is an article having a metallic luster and the background color in the inspection object image is black. In this case, in the inspection target image, the brightness of the pixels in the region corresponding to the inspection target is high and the brightness of the pixels belonging to the background region is low, as in the example shown in FIG. In the inspection object image, the first partial region is designated in the part where the inspection object is reflected, and the first frame is designated so as to surround the inspection object in the background portion around the inspection object. If so, the brightness of the pixels belonging to the first partial region becomes high, and the brightness of the pixels belonging to the first frame becomes low. In the watershed method, the brightness is regarded as the water level, and as shown in FIG. 4, the brightness is increased by a certain amount for the pixels with low brightness, and the brightness is decreased by a certain amount for the pixels with high brightness. Then, as shown in FIG. 5, the portion where the brightness of both coincides includes the boundary between the region to which the pixel having low brightness belongs and the region to which the pixel having high brightness belongs, that is, the first partial region and includes the first frame. Not specified as a contour candidate for the extraction target area.

The contour candidates identified by the contour candidate specifying unit 120 are displayed on the display device 30 by the display control unit 100. When the user is dissatisfied with the specified contour candidate, such as when the contour candidate displayed on the display device 30 includes a subject other than the inspection target, the user sets the first partial area. An operation of invalidating at least one of the first frames can be performed on the input device 40. Specific examples of the operation of invalidating the first partial area and the first frame include an operation of pressing a predetermined key such as the X key while clicking with the mouse.

The invalidation unit 130 invalidates the designation of at least one of the first partial area and the first frame recognized by the recognition unit 110 according to the operation content data given from the input device 40. The user can make the contour candidate specifying unit 120 specify a desired contour candidate by redesignating the first partial region or the first frame.

The update unit 140 generates a second partial region within the contour candidate and a second frame outside the contour candidate and surrounding the contour candidate based on the contour candidate identified by the contour candidate specifying unit 120. More specifically, the updating unit 140 generates a region surrounded by the contour obtained by contracting the contour candidate identified by the contour candidate specifying unit 120 at a predetermined reduction rate as the second partial region, and the contour The specified contour candidate of the candidate identification unit 120 is expanded at a predetermined enlargement ratio to generate a second frame. The reduction ratio when the second partial region is generated and the enlargement ratio when the second frame is generated may be determined according to the manufacturing variation regarding the size of the inspection object.

The area specifying unit 150 specifies the extraction target area so as to include the second partial area and not to include the second frame based on the second partial area and the second frame generated by the updating unit 140. More specifically, the region specifying unit 150 specifies the outline of the extraction target region by the above-mentioned watershed method so as to include the second partial region and not to include the second frame. Then, the area specifying unit 150 is a mask for extracting the inspection target area from the inspection target image by using the data representing the position, size, and shape of the extraction target area specified in the above procedure in the inspection target image. It is written in the storage device 50 as information.

The appearance inspection unit 160 identifies an inspection target area in the inspection target image with reference to the mask information stored in the storage device 50, and performs image processing for the appearance inspection on the inspection target area. As for this visual inspection algorithm, an existing algorithm may be used as appropriate.

In the inspection device 1 of the present embodiment, the processing device 10 executes the image processing method shown in the flowchart of FIG. 6 according to the control program 510. As shown in FIG. 6, this image processing method includes a recognition process SA100, a contour candidate identification process SA110, an update process SA120, and an area identification process SA130. Hereinafter, as shown in FIG. 7, the images from which the inspection target area is extracted are the article A which is the inspection target and the articles B and C which are not the inspection targets but have a metallic luster like the article A. The image G01 in which the background is black is the image G01, and the user does not accurately follow the shape of the article A with respect to the image G01, but the first partial region R01 and the first frame W01. The processing contents of each processing will be described by taking as an example the case where each of the above is roughly specified by dragging the mouse.

In the recognition process SA100, the processing device 10 functions as the recognition unit 110 and recognizes the first subregion designated by the user and the first frame designated by the user. In the recognition process SA100 of this operation example, the first partial region R01 and the first frame W01 shown in FIG. 8 are recognized based on the operation content data given from the input device 40.

In the contour candidate identification process SA110, the processing device 10 functions as the contour candidate identification unit 120, and the first portion region recognized by the recognition process SA100 and the first frame recognized by the recognition unit 110 are used. The contour candidates of the extraction target area are specified so as to include the partial area and not include the first frame. When the above-mentioned watershed method is applied to the image G01 shown in FIG. 7, the contour of the article A, the contour of the article B, and the contour of the article C are specified as the contours of the low-luminance region and the high-luminance region. .. Of these three contours, the first partial region R01 is included. Moreover, only the outline of the article A does not include the first frame W01. Therefore, in the contour candidate identification process SA110, as shown in FIG. 9, the contour candidate Z01 which is substantially the same as the contour of the article A is specified.

In the update processing SA120, the processing device 10 functions as an update unit 140, and based on the contour candidate identified by the contour candidate identification processing SA110, the second partial region in the contour candidate and the contour outside the contour candidate and outside the contour candidate. Generate a second frame that surrounds the candidate. In this operation example, the contour candidate Z01 is specified by the contour candidate identification process SA110 as shown in FIG. In the update process SA120, as shown in FIG. 10, a second partial region R02 having a contour obtained by contracting the contour candidate Z01 and a second frame W02 obtained by expanding the contour candidate Z01 are generated.

In the area specifying process SA130, the processing device 10 functions as the area specifying unit 150, and includes the second partial area and includes the second frame based on the second partial area and the second frame generated by the update process SA120. Specify the extraction target area so that it is not included. In this operation example, the second partial region R02 and the second frame W02 are generated in the update process SA120 as shown in FIG. Therefore, in the area identification process SA130, an extraction target area having substantially the same contour as the contour of the member A is specified as in the contour candidate Z01 shown in FIG. 9 above.

After that, in the visual inspection of the article A, the inspection device 1 extracts the inspection target area according to the mask information stored in the storage device 50 every time the inspection target image is input, and the visual inspection is performed.

Since the contour of the extraction target area specified by the area identification process SA130 in the operation example described above is substantially the same as the contour candidate specified by the contour candidate identification process SA110, the update process SA120 and the area identification process SA130 Seems to be unnecessary processing. The reason why the update process SA120 and the area identification process SA130 are provided in this embodiment is as follows.

It is assumed that the image that is the extraction source of the extraction target area is the image G02 shown in FIG. The image G02 shown in FIG. 11 shows the article A, the article B, and the article C as in the image G01 of FIG. 7, but differs from the image G01 in that the article B is shown close to the article A. When the user roughly specifies the first partial region R01 and the first frame W01 with respect to the image G02 as shown in FIG. 11, the contour candidate Z01 ′ including the contour of the article B is generated as shown in FIG. It may be specified by the contour candidate identification process SA110. If the update process SA120 and the area identification process SA130 are not performed, the contour candidate Z01'shown in FIG. 12 will be the contour of the extraction target region, and the defect due to the difference in background or the like will be included as much as the contour of the article B is included. False positives are more likely to occur. On the other hand, when the update process SA120 and the area identification process SA130 are performed after the contour candidate identification process SA110, the second partial area R02'and the second frame W02'shown in FIG. 13 are in the update process SA120. In the generated area identification process SA130, an extraction target area having substantially the same contour as the contour of the article A is specified as in the above operation example. That is, in the update processing SA120 and the area identification processing SA130, the user can use the first partial area and the area identification process SA130 in a situation where the image of the extraction source of the extraction target area varies, such as when the position of the article is slightly deviated. Even if the first frame is roughly specified, it is provided so that the extraction target area can be extracted with high accuracy. This is the reason why the update process SA120 and the area identification process SA130 are provided.

As described above, the inspection device 1 has a recognition unit 110, a contour candidate identification unit 120, an update unit 140, and an area identification unit 150. The recognition unit 110 recognizes the first partial area in the extraction target area designated by the user and the first frame outside the extraction target area designated by the user and surrounding the extraction target area. The contour candidate specifying unit 120 specifies the contour candidate of the extraction target region based on the first partial region and the first frame so as to include the first partial region and not include the first frame. Based on the contour candidate, the update unit 140 generates a second partial region within the contour candidate and a second frame outside the contour candidate and surrounding the contour candidate. Then, the region specifying unit 150 specifies the extraction target region based on the second partial region and the second frame so as to include the second partial region and not to include the second frame. Therefore, even if the user roughly specifies the first subregion and the first frame, it is possible to accurately extract the extraction target region having substantially the same contour as the contour of the inspection target, and the extraction target region. It is possible to prevent the occurrence of false detection of defects due to the inclusion of extra background in.

Further, the inspection device 1 causes the display device 30 to display the image of the inspection object, the contour candidate specified by the contour candidate identification unit 120, the first partial region and the first frame designated by the user, and the display control unit 100. Has. Therefore, the user of the inspection device 1 can specify the first partial area and the first frame while visually checking the image of the inspection object displayed on the display device 30. In addition, the user can visually confirm the contour candidates specified according to the first subregion and the first frame designated by the user.

Further, the inspection device 1 further includes an invalidation unit 130 that invalidates the designation of at least one of the first partial region and the first frame recognized by the recognition unit 110. Therefore, when the user is dissatisfied with the contour candidate specified by the contour candidate specifying unit 120, at least one of the first partial region and the first frame is invalidated until a desired contour candidate is obtained. Redesignation and identification of contour candidates by the contour candidate identification unit 120 can be repeated.

B. Other Embodiments The following modifications may be appropriately combined with the above embodiments.

(1) The inspection device 1 of the above embodiment includes an image pickup device 20, a display device 30, an input device 40, and a storage device 50 in addition to the processing device 10. However, one or more of the image pickup device 20, the display device 30, the input device 40, and the storage device 50, or all of them are connected to the inspection device 1 via a telecommunication line such as a LAN (Local Area Network) or the Internet. May be done.

(2) Each part of the display control unit 100, the recognition unit 110, the contour candidate identification unit 120, the invalidation unit 130, the update unit 140, the area identification unit 150, and the appearance inspection unit 160 in the above embodiment is a control program that is software. It was a software module realized by a computer according to 510. However, each part of the display control unit 100, the recognition unit 110, the contour candidate identification unit 120, the invalidation unit 130, the update unit 140, the area identification unit 150, and the appearance inspection unit 160 is realized by hardware such as ASIC, and these hardwares are realized. The inspection device 1 may be configured by combining hardware.

(3) The update unit 140 of the above embodiment shrinks the contour candidate identified by the contour candidate identification unit 120 to generate the second partial region, but the first partial region recognized by the recognition unit 110, that is, the use The second subregion may be generated so as to include the first subregion specified by the person. The reason is as follows.

For example, it is assumed that the article A in FIG. 14 is an object to be inspected, and the portion indicated by the hatching of the diagonal line in FIG. 14 is a particularly noteworthy part in the inspection. In such a case, when the contour candidate Z01 is contracted to generate the second partial region R02 as shown in FIG. 10 above, as shown in FIG. 14, a part of the portion notable in the inspection is the second partial region R02. It can happen that it sticks out. As described above, when a part of the portion to be noted in the inspection protrudes from the second partial region R02, there is a possibility that the extraction target region not including a part of the portion to be noted in the inspection is specified.

On the other hand, as shown in FIG. 15, the user is made to specify the first subregion R01 ′ so as to include all the points of interest in the inspection, and the second subregion R02 ″ is shown in FIG. If the update unit 140 is made to generate the second partial region including the first partial region R01', the extraction target region includes all the points of interest, and the inspection omission of the relevant portion is ensured. It becomes possible to avoid it. As a mode of generating the second partial region including the first partial region designated by the user, as shown in FIG. 17, the first partial region R01'designated by the user and the contour candidate Z01 are contracted. It is conceivable that the region obtained by superimposing the obtained second partial region R02 is referred to as the second partial region R02 ″.

(4) In the above embodiment, the user is made to specify the first partial area and the first frame by operating the inspection object image displayed on the display device 30. However, on the recording paper on which the image of the inspection object is printed, the user is made to add a mark indicating the first partial area and a mark indicating the first frame in different notation modes, and the recording paper on which these marks are added is added. May be read by a scanner, and the recognition unit 110 may detect the first partial region and the first frame based on each of the above marks by image processing on the read image. In this embodiment, the display of the inspection target image, the extraction target region, the first partial region, and the first frame on the display device 30 is not essential, so that the display control unit 100 can be omitted. Further, since the invalidation of the first partial area and the invalidation of the first frame designated by the user are not essential, the invalidation unit 130 can also be omitted.

(5) Although the inspection device 1 of the above embodiment executes the extraction of the inspection target area and the visual inspection, each of the extraction of the inspection target area and the visual inspection may be executed by another device. Specifically, an inspection device in which another device functioning as an appearance inspection unit 160 is connected to an image processing device configured by combining a recognition unit 110, a contour candidate identification unit 120, an update unit 140, and an area identification unit 150. 1 may be configured. Further, the image processing apparatus may be manufactured or sold as a single unit. This is because the image processing device can be used not only for extracting the inspection target area in the visual inspection but also for other purposes such as extracting the face area for face recognition.

(6) In the above embodiment, each process of the image processing method shown in FIG. 6 is executed by the processing device 10. However, each of the four processes in FIG. 6 may be performed by a different device. In short, it is an image processing method that specifies an extraction target area for extracting an image from the input image, and is an image processing method that includes the following recognition processing, contour candidate identification processing, update processing, and area identification processing. In the recognition process, the first sub-region within the extraction target area designated by the user and the first frame outside the extraction target area designated by the user and surrounding the extraction target area are recognized. In the contour specifying process, the contour candidates of the extraction target region are specified based on the first partial region and the first frame so as to include the first partial region and not include the first frame. In the update process, a second subregion within the contour candidate and a second frame outside the contour candidate and surrounding the contour candidate are generated based on the contour candidate. Then, in the area identification process, the extraction target area is specified based on the second partial area and the second frame so as to include the second partial area and not include the second frame.

(7) In the above embodiment, the control program 510 that causes the processing device 10 to function as the recognition unit 110, the contour candidate identification unit 120, the update unit 140, and the area identification unit 150 is stored in the storage device 50 in advance. It had been. However, a program that causes a general computer to function as a recognition unit 110, a contour candidate identification unit 120, an update unit 140, and an area identification unit 150 may be manufactured and distributed as a single unit. This is because by operating a general computer according to this program, the computer can function as the image processing device. As a specific distribution method of the above program, the above program is written and distributed on a computer-readable recording medium such as a flash ROM (Read Only Memory), or is distributed by downloading via a telecommunication line such as the Internet. The mode of doing so is conceivable.

(8) In the above embodiment, the user has designated the first partial region and the first frame, but the inspection device 1 recognizes the inspection object image, and the inspection device 1 is based on the inspection object image. May automatically specify the first subregion and the first frame. Further, the inspection robot or the like may specify the first subregion and the first frame instead of the user.
The present disclosure is not limited to the above-described embodiment, and can be widely applied without departing from the gist of the present disclosure.

1 ... Inspection device, 10 ... Processing device, 100 ... Display control unit, 110 ... Recognition unit, 120 ... Contour candidate identification unit, 130 ... Invalidation unit, 140 ... Update unit, 150 ... Area identification unit, 160 ... Appearance inspection unit , 20 ... Imaging device, 30 ... Display device, 40 ... Input device, 50 ... Storage device, 510 ... Control program, 60 ... Bus.

Claims (8)

An image processing device that specifies an extraction target area for extracting an image from an input image.
A first partial region within the extraction target region, a recognition unit outside the extraction target region and recognizing a first frame surrounding the extraction target region, and a recognition unit.
A contour candidate specifying unit that specifies contour candidates of the extraction target region so as to include the first partial region and not include the first frame based on the first partial region and the first frame.
Based on the contour candidate, a second partial region in the contour candidate, an update unit outside the contour candidate and generating a second frame surrounding the contour candidate, and
A region specifying portion that specifies the extraction target region so as to include the second partial region and not include the second frame based on the second partial region and the second frame.
An image processing device comprising. The first aspect of claim 1, wherein the display control unit has a display control unit for displaying the image, the first partial region, the first frame, and the contour candidate identified by the contour candidate specifying unit on a display device. Image processing equipment. The image processing apparatus according to claim 2, further comprising an invalidation unit that invalidates at least one of the first region and the first frame. The update unit contracts the contour candidate identified by the contour candidate specifying unit to generate the second partial region, and expands the contour candidate identified by the contour candidate specifying unit to generate the second frame. The image processing apparatus according to any one of claims 1 to 3, wherein the image processing apparatus is used. The image processing apparatus according to any one of claims 1 to 3, wherein the updating unit generates the second partial region so as to include the first partial region. In order to extract the extraction target area corresponding to the inspection target object in the input image, the first partial area in the extraction target area and the first frame outside the extraction target area and surrounding the extraction target area are recognized. Recognition part and
A contour candidate specifying unit that specifies contour candidates of the extraction target region so as to include the first partial region and not include the first frame based on the first partial region and the first frame.
Based on the contour candidate, a second partial region in the contour candidate, an update unit outside the contour candidate and generating a second frame surrounding the contour candidate, and
A region specifying portion that specifies the extraction target region so as to include the second partial region and not include the second frame based on the second partial region and the second frame.
Of the input images, an appearance inspection unit that inspects the appearance of the inspection object by referring to the image of the extraction target area identified by the area identification unit, and
Inspection device equipped with. An image processing method for specifying an extraction target area for extracting an image from an input image, which is a first partial area within the extraction target area and a first frame outside the extraction target area and surrounding the extraction target area. Is recognized, and based on the first partial region and the first frame, contour candidates of the extraction target region are specified so as to include the first partial region and not include the first frame, and the contour candidates are selected as the contour candidates. Based on this, a second partial region within the contour candidate and a second frame outside the contour candidate that surrounds the contour candidate are generated, and based on the second partial region and the second frame, the second frame is generated. An image processing method for specifying the extraction target area so as to include a partial area and not include the second frame. Computer,
In order to specify the extraction target area of the input image to be extracted, the first partial area in the extraction target area and the first frame outside the extraction target area and surrounding the extraction target area are recognized. Recognition part and
A contour candidate specifying unit that specifies contour candidates of the extraction target region so as to include the first partial region and not include the first frame based on the first partial region and the first frame.
Based on the contour candidate, a second partial region in the contour candidate, an update unit outside the contour candidate and generating a second frame surrounding the contour candidate, and
A region specifying portion that specifies the extraction target region so as to include the second partial region and not include the second frame based on the second partial region and the second frame.
A program that works.

Images