WO2020145164A1

WO2020145164A1 - Image processing device, image processing program, and image processing method

Info

Publication number: WO2020145164A1
Application number: PCT/JP2019/050990
Authority: WO
Inventors: 加藤　豊
Original assignee: オムロン株式会社
Priority date: 2019-01-09
Filing date: 2019-12-25
Publication date: 2020-07-16
Also published as: JP7049587B2; DE112019006614T5; JP2020112938A; CN112689851A

Abstract

Provided are an image processing device, an image processing program, and an image processing method, with which values for a plurality of parameters can be set efficiently. An image processing device 50, which is for carrying out a process for a processing image on the basis of a plurality of parameters, is equipped with: a prior knowledge generation unit 44, which uses a plurality of learning images, an evaluation item corresponding to the plurality of learning images, and a target value for the evaluation item to generate prior knowledge 34 related to the parameters; and a setting unit 42, which uses the prior knowledge 34, the plurality of processing images, the evaluation item corresponding to the plurality of processing images, and the target value for the evaluation item to set a value for the parameters.

Description

Image processing apparatus, image processing program, and image processing method

The present invention relates to an image processing device, an image processing program, and an image processing method.

Conventionally, as a parameter determination support device, a parameter determination support for an image processing device that obtains a processing result by performing processing using a set of predetermined parameters on image data obtained by imaging a measurement target object. The apparatus is an apparatus for receiving image data and an expected result associated with the image data, and a candidate generation for generating a plurality of parameter candidates in which at least one parameter value included in the parameter set is different from each other. Means for acquiring a plurality of processing results using each of a plurality of parameter candidates for the image data, and a plurality of processing results by comparing each of the plurality of processing results with the corresponding expected result. It is known that an evaluation unit that generates an evaluation result for each result and an output unit that outputs the evaluation result for each of a plurality of parameter candidates are provided (see Patent Document 1). This parameter determination support device is a set of parameters set in an image processing device that obtains a processing result by performing processing using a predetermined set of parameters on image data obtained by imaging a measurement target object. Can be determined more quickly and easily.

JP, 2010-191939, A

On the other hand, in the case of setting the values of a plurality of parameters, when trying to set the value of an appropriate parameter, for example, all combinations of the values of the plurality of parameters are evaluated, and the combination of the values of the parameters having the highest evaluation value is set. Had to choose. Therefore, it takes a very long time to set the parameter value, which is not efficient.

Therefore, an object of the present invention is to provide an image processing device, an image processing program, and an image processing method capable of efficiently setting the values of a plurality of parameters.

An image processing apparatus according to an aspect of the present invention is an image processing apparatus that performs processing on a processing image based on a plurality of parameters, and includes a plurality of learning images and evaluation items corresponding to the plurality of learning images. And a target value of the evaluation item, an a priori knowledge generation unit that generates a priori knowledge about the parameter, an a priori knowledge, a plurality of processing images, an evaluation item corresponding to the plurality of processing images, and an evaluation item of the evaluation item. And a setting unit that sets the value of the parameter based on the target value.

According to this aspect, the value of the parameter is set based on the prior knowledge about the parameter, the plurality of processing images, the evaluation item corresponding to the plurality of processing images, and the target value of the evaluation item. This makes it possible, for example, to search for a value in an appropriate range with an appropriate range of change for a parameter, so that an appropriate parameter value can be set in a shorter time than when there is no prior knowledge. can do. Therefore, the values of the plurality of parameters can be efficiently set.

In the above-described aspect, the prior knowledge generation unit uses an evaluation function defined by the evaluation item and the target value of the evaluation item, and an evaluation unit for obtaining an evaluation value of the evaluation function corresponding to the value of the parameter and the evaluation unit. The a priori knowledge may be generated based on the plurality of evaluation values obtained in step 1, and the setting unit may set the value of the parameter so that the evaluation value becomes high.

According to this aspect, the value of the parameter is set so that the evaluation value becomes high. This makes it possible to set the value of the parameter close to or satisfying the target value of the evaluation item. Further, using the evaluation function defined by the evaluation item and the target value of the evaluation item, the evaluation means for obtaining the evaluation value of the evaluation function corresponding to the value of the parameter, and a plurality of evaluation values obtained by the evaluation means Prior knowledge is generated based on With this, for example, for a parameter suitable for increasing the evaluation value, it is possible to search for a value in an appropriate range with an appropriate change range, and thus in a short time as compared with the case without prior knowledge, It is possible to set the value of a parameter having a higher evaluation value.

In the above-mentioned aspect, the evaluation means may be an evaluation value map generated using an evaluation function and showing an evaluation value for each parameter value in the evaluation item.

According to this aspect, the evaluation means is an evaluation value map which is generated by using the evaluation function and indicates the evaluation value for each value of the parameter in the evaluation item. This makes it possible to easily realize the prior knowledge generation unit that generates the prior knowledge.

In the above-mentioned aspect, the prior knowledge generation unit generates the prior knowledge for each type of processing or each business type of the processing image object, and selects the selection unit for selecting the prior knowledge based on the designated type or business type. You may further prepare.

According to this aspect, prior knowledge is generated for each type of processing or each industry of the object of the processing image, and prior knowledge is selected based on the designated type or industry. Thereby, it is possible to obtain common prior knowledge for each type of processing or the type of business of the object.

In the above-mentioned aspect, the prior knowledge may include at least one of the degree of influence of the parameter on the evaluation item, the degree of dependence between the parameters, the range of change of the parameter value, and the range of change of the parameter value.

According to this aspect, the prior knowledge includes at least one of the degree of influence of the parameter on the evaluation item, the degree of dependence between the parameters, the range of change of the value of the parameter, and the range of change of the value of the parameter. This makes it possible to easily realize a setting unit that sets a parameter having a higher evaluation value in a short time.

In the above-described aspect, the setting unit may set a parameter having a highest evaluation value within the designated set time range.

According to this aspect, a value having the highest evaluation value within the specified set time range is set as the parameter. As a result, the value of the parameter having a higher evaluation value can be set even within the limited setting time.

In the above-described aspect, the setting unit may further include an extraction unit that extracts at least one of the plurality of parameters based on prior knowledge, and the setting unit may set the value of the extracted parameter.

According to this aspect, at least one of the plurality of parameters is extracted based on the prior knowledge, and the value of the extracted parameter is set. With this, it is possible to set the value of the parameter that can increase the evaluation value among the plurality of parameters.

An image processing program according to another aspect of the present invention is an image processing program that causes a computer to perform processing on a processing image based on a plurality of parameters, and includes a plurality of learning images and the plurality of learning images. A prior knowledge generation step of generating prior knowledge about parameters using an evaluation item corresponding to the learning image and a target value of the evaluation item, corresponding to the prior knowledge, a plurality of processing images, and the plurality of processing images. The setting step of setting the value of the parameter based on the evaluation item to be performed and the target value of the evaluation item.

According to this aspect, the value of the parameter is set based on the prior knowledge about the parameter, the plurality of processing images, the evaluation items corresponding to the plurality of processing images, and the target value of the evaluation item. This makes it possible, for example, to search for a value in an appropriate range with an appropriate range of change for a parameter, so that an appropriate parameter value can be set in a shorter time than when there is no prior knowledge. can do. Therefore, the values of the plurality of parameters can be efficiently set.

In the above-described aspect, the a priori knowledge generation step uses an evaluation function defined by the evaluation item and the target value of the evaluation item, and an evaluation unit for obtaining an evaluation value of the evaluation function corresponding to the value of the parameter and the evaluation unit. The method may include generating a priori knowledge based on the plurality of evaluation values obtained in 1., and the setting step may include setting the value of the parameter so that the evaluation value becomes high.

According to this aspect, the evaluation means is an evaluation value map which is generated by using the evaluation function and indicates the evaluation value for each value of the parameter in the evaluation item. Thereby, the prior knowledge generating step for generating the prior knowledge can be easily realized.

In the above-mentioned aspect, the a priori knowledge generation step includes generating a priori knowledge for each type of processing or each business type of the object of the processing image, and selects the prior knowledge based on the specified type or business type. It may further include a selection step.

According to this aspect, a priori knowledge is generated for each type of processing or for each business type of the target of the processing image, and the evaluation value map is selected for each type of processing or each business type of the target. Thereby, it is possible to obtain common prior knowledge for each type of processing or the type of business of the object.

In the above-mentioned aspect, the setting step may include setting a parameter having a highest evaluation value within the designated set time range.

The above-described aspect may further include an extraction step of extracting at least one of the plurality of parameters based on prior knowledge, and the setting step may set the value of the extracted parameter.

An image processing method according to another aspect of the present invention is an image processing method for performing processing on a processing image based on a plurality of parameters, and corresponds to a plurality of learning images and the plurality of learning images. Using the evaluation item and the target value of the evaluation item, a prior knowledge generation step of generating prior knowledge about the parameter, the prior knowledge, the plurality of processing images, and the evaluation items corresponding to the plurality of processing images A setting step of setting the value of the parameter based on the target value of the evaluation item.

According to the present invention, it is possible to efficiently set the values of multiple parameters.

FIG. 1 is a configuration diagram illustrating a schematic configuration of an image processing system according to an embodiment. FIG. 2 is a diagram illustrating the evaluation function shown in FIG. FIG. 3 is a diagram showing an example of the evaluation value map shown in FIG. FIG. 4 is a diagram showing an example of the evaluation value map shown in FIG. FIG. 5 is a diagram illustrating the evaluation value map shown in FIG. 1. FIG. 6 is a diagram showing an example of the learning data set shown in FIG. FIG. 7 is a diagram showing an example of the evaluation value map shown in FIG. FIG. 8 is a flowchart showing a schematic operation for setting the parameters of the image processing apparatus according to the embodiment.

An embodiment of the present invention will be described below. In the following description of the drawings, the same or similar parts are denoted by the same or similar reference numerals. However, the drawings are schematic. Therefore, specific dimensions and the like should be determined in light of the following description. Further, it is needless to say that the drawings include portions in which dimensional relationships and ratios are different from each other. Furthermore, the technical scope of the present invention should not be understood as being limited to the embodiment.

First, an example of the configuration of an image processing apparatus according to an embodiment of the present invention will be described with reference to FIGS. 1 to 7. FIG. 1 is a configuration diagram illustrating a schematic configuration of an image processing system 100 according to an embodiment. FIG. 2 is a diagram illustrating the evaluation function 31 shown in FIG. FIG. 3 is a diagram illustrating the evaluation value map 32 shown in FIG. FIG. 4 is a diagram illustrating the evaluation value map 32 shown in FIG. FIG. 5 is a diagram illustrating the evaluation value map 32 shown in FIG. FIG. 6 is a diagram illustrating the learning data set 33 illustrated in FIG. 1. FIG. 7 is a diagram illustrating the evaluation value map 32 shown in FIG.

As shown in FIG. 1, the image processing system 100 includes a server SV, a network NW, an image processing device 50, an imaging device 60, an input device 70, an output device 80, and a line device 90.

The image processing device 50 and the server SV are communicably connected to each other via the network NW. The network NW is, for example, the Internet, LAN (Local Area Network), leased line, telephone line, corporate network, Bluetooth (registered trademark), Wi-Fi (Wireless Fidelity), mobile communication network, other communication lines, or These are combinations and the like. Further, the connection between the image processing apparatus 50 and the server SV and the network NW may be wired or wireless.

In the present embodiment, an example in which the number of networks NW included in the image processing system 100 is shown in FIG. 1, but the present invention is not limited to this. The number of networks included in the image processing system 100 may be two or more, for example.

The server SV is a computer that functions as a host controller of the image processing device 50. The server SV is configured to include a controller such as a PLC (Programmable Logic Controller), for example.

The image processing device 50 processes a processing image based on a plurality of parameters. More specifically, the image processing device 50 inspects the processing image of the target object, determines the position from the processing image of the target object, and determines the processing image of the target object based on the plurality of parameters. It is to collate characters from.

In the present embodiment, for example, the object on the line device 90 is photographed by the imaging device 60, and the image processing device 50 inspects the image for processing the photographed object. In this case, the plurality of parameters are, for example, the size of the kernel or the filter when smoothing the processing image, the threshold value when binarizing the processing image, and the like.

The image processing device 50 is, for example, an information processing device such as a computer. The image processing apparatus 50 includes a communication unit 10, an input/output interface (hereinafter, referred to as “input/output I/F”) 20, a storage unit 30, and a control unit 40. Further, the image processing device 50 further includes a bus 51 configured to transmit signals and data between the respective units of the image processing device 50.

The communication unit 10 is for communicating (transmitting/receiving) data via the network NW. The communication unit 10 is configured to be able to communicate with the server SV connected to the network NW via the network NW based on one or a plurality of predetermined communication methods.

The input/output I/F 20 is an interface between the image processing device 50 and an external device. The input/output I/F 20 is configured to exchange data and signals with external devices. Further, the input/output I/F 20 is configured to control communication with an external device. The input/output I/F 20 includes connection terminals of standardized interfaces such as USB (Universal Serial Bus), HDMI (registered trademark) (High-Definition Multimedia Interface), and IEEE 1394. The input/output I/F 20 is connected to the imaging device 60, the input device 70, the output device 80, and the line device 90.

The storage unit 30 is configured to store programs and data. The storage unit 30 includes, for example, a hard disk drive, a solid state drive, and the like. The storage unit 30 stores various programs executed by the control unit 40, data necessary for executing the programs, and the like in advance.

The storage unit 30 also stores in advance an evaluation function 31, an evaluation value map 32, a learning data set 33, prior knowledge 34, a search history 35, and a calculation result 36.

The evaluation function 31 is defined by the evaluation item and the target value of the evaluation item. The evaluation item is an item defined corresponding to a plurality of images. In the present embodiment, the evaluation items correspond to the plurality of learning images included in the learning data set 33. As shown in FIG. 2, the evaluation items are, for example, items such as "processing time", "overlooking rate", "missing rate", "position accuracy", "posture accuracy", and "dimensional accuracy". A target value is set for each evaluation item. In the example shown in FIG. 2, the target value includes two target values, a mast target value and a want target value.

Each target value is input by the user (user) operating the input device 70. The evaluation function 31 is, for example, 60 points or 60% when the mast target values of all the evaluation items are satisfied, and is 90 points when the want target values of all the evaluation items are satisfied, depending on the parameter values of the image processing device 50. The quantity is normalized to be 90%.

The evaluation value map 32 is generated using the evaluation function 31 shown in FIG. 2, and is generated for each evaluation item of the evaluation function 31. The evaluation value map 32 shows the evaluation value of the evaluation function (hereinafter, simply referred to as “evaluation value”) for each value of the parameter of the image processing device 50 in the evaluation item.

For example, the evaluation value map 32 shown in FIG. 3 maps the evaluation values of the parameters p1, p2, and p3 in the parameter space in the “processing time” which is one of the evaluation items of the evaluation function 31.

Further, the evaluation value map 32 shown in FIG. 4 is a map obtained by mapping the evaluation values of the parameters p1, p2, and p3 in the parameter space in the “oversight rate” which is another evaluation item of the evaluation function 31.

The evaluation value map 32 is not limited to the one in which evaluation values in a plurality of parameter spaces are mapped for each evaluation item. For example, as shown in FIG. 5, in the evaluation item “processing time”, an evaluation value for the value of the parameter p1 may be mapped.

The prior knowledge 34 is generated based on the evaluation value map 32 as shown in FIGS. 3 to 5, for example. The prior knowledge 34 is related to the parameters, and is, for example, the degree of influence of the parameters on the evaluation item, the degree of dependence between the parameters, the step size of the parameters, the range of parameter changes, and the like. FIG. 5 shows an example of obtaining the change width of the parameter p1 as the prior knowledge 34.

When the degree of influence of the parameter on the evaluation item is obtained as the prior knowledge 34, when the evaluation value of a certain evaluation item is low, the parameter whose value should be changed to increase the evaluation value of the evaluation item is known. Further, when the degree of dependence between parameters is obtained as the prior knowledge 34, when changing a certain parameter, the parameter to be changed is known at the same time. Further, when the parameter change width is obtained as the prior knowledge 34, an appropriate change width in the value of the parameter can be known when searching for a value having a higher evaluation value for a certain parameter. Further, when the parameter change range is obtained as the prior knowledge 34, an appropriate change range in the value of the parameter can be known when searching for a value having a higher evaluation value for a certain parameter.

In the present embodiment, an example in which the prior knowledge 34 is generated based on the evaluation value map 32 has been shown, but the present invention is not limited to this. For example, a method of performing some kind of trial experiment on a problem prepared in advance and grasping the tendency of parameters can be considered. In this case, a priori knowledge may be obtained from the tendency of the parameters.

The evaluation value map 32 is generated using the learning data set 33. As shown in FIG. 6, the learning data set 33 includes a plurality of learning data sets, and the evaluation value map 32 is generated using at least one learning data set in the learning data set 33. ..

Each learning data includes, for example, a plurality of learning images, a type of processing to be performed on the plurality of learning images, the number of learning images, and a label given to each learning image. Contains. The label is related information according to the type of processing, that is, an annotation. For example, when the processing type is “inspection”, the label is “OK” or “NG”, and when the processing type is “positioning”, the label is a position (x, y) of xy coordinates with a certain pixel as a reference or origin, When the processing type is "character collation", the label is a date such as year and month.

The evaluation value map 32 may be generated for each evaluation item, for each type of processing performed on the processing image, or for each business type of the target object shown in the processing image. For example, as shown in FIG. 7, the evaluation value map 32 uses the learning data whose processing type is “inspection”, and maps the evaluation values in the parameter space of p1, p2, and p3 for the evaluation item “processing time”. It was done.

The evaluation value map 32 stored in the storage unit 30 is not limited to the one generated by the image processing device 50. For example, the server SV shown in FIG. 1 may create the evaluation value map. In this case, the server SV stores the evaluation function and the learning data set in advance, or acquires the evaluation function and the learning data set from another device via the network NW. Then, the evaluation value map generated by the server SV is transmitted to the image processing apparatus 50 via the network NW and stored in the storage unit 30.

The structures and formats of the evaluation function 31, the evaluation value map 32, the learning data set 33, the prior knowledge 34, the search history 35, and the calculation result 36 are not limited to the examples described above. For example, the evaluation function 31, the evaluation value map 32, the learning data set 33, the prior knowledge 34, the search history 35, and the calculation result 36 may each be simple data or a database. If at least a part of the evaluation function 31, the evaluation value map 32, the learning data set 33, the prior knowledge 34, the search history 35, and the calculation result 36 is a database, normalization is performed and the data is grouped into groups. May be subdivided.

Returning to the description of FIG. 1, the control unit 40 is configured to control the operation of each unit of the image processing apparatus 50, such as the communication unit 10, the input/output I/F 20, the storage unit 30, and the like. Further, the control unit 40 is configured to realize each function described below by executing a program stored in the storage unit 30 or the like. The control unit 40 is configured to include a processor such as a CPU (Central Processing Unit), a memory such as a ROM (Read Only Memory), a RAM (Random Access Memory), and a buffer storage device such as a buffer. The details of the control unit 40 will be described later.

The imaging device 60 is configured to capture an image and record it as data. The imaging device 60 is a digital camera, and is configured to include, for example, optical system components such as a lens and electronic system components such as an imaging element (light receiving element). The optical system component may include a plurality of lenses. Moreover, the electronic component may include a light emitting device such as a flash. The imaging device 60 is arranged above the line device 90, photographs the object on the line device 90, and outputs the photographed image. The image output by the imaging device 60 is input to the image processing device 50 via the input/output I/F 20. The control unit 40 performs necessary processing on the image input from the imaging device 60 to generate a processing image file, and stores the generated processing image file in the storage unit 30.

The input device 70 is configured so that information can be input by a user (user) operation. The input device 70 includes, for example, a keyboard, a keypad, a mouse, a trackball, a touch panel, a microphone, and the like. For example, when the user operates a keyboard, a keypad, a mouse, a trackball, a touch panel, a microphone or the like (including a voice operation using a microphone), the input device 70 outputs data or a signal corresponding to the operation. Output. The data or signal output by the input device 70 is input to the image processing device 50 via the input/output I/F 20. The control unit 40 generates data based on this data or signal, so that information can be input to the image processing device 50.

The output device 80 is configured to output information. The output device 80 is configured to include, for example, a display device such as a liquid crystal display, an EL (Electro Luminescence) display, and a plasma display, and an audio device such as a speaker.

The output device 80 displays information such as text such as characters, numbers, and symbols, images, and moving images on the display device among the information input from the image processing device 50 via the input/output I/F 20, and the audio information. By outputting from the speaker, it becomes possible to output information.

The line device 90 is configured to convey an object. The line device 90 is configured to include a conveying unit such as a belt conveyor. The line device 90 moves, stops, or excludes an object on the line device 90, for example, based on a control signal input from the image processing device 50 via the input/output I/F 10. Will be possible.

The control unit 40 of the image processing device 50 includes, for example, a selection unit 41, a setting unit 42, an extraction unit 43, and a prior knowledge generation unit 44 as its functional configuration.

The prior knowledge generation unit 44 uses the plurality of learning images included in the learning data of the learning data set 33, the evaluation items corresponding to the plurality of learning images, and the target values of the evaluation items, and It is configured to generate the prior knowledge 34 described above. The prior knowledge generation unit 44 stores the generated prior knowledge 34 in the storage unit 30.

More specifically, the prior knowledge generation unit 44 uses an evaluation function 31 defined by the evaluation item and the target value of the evaluation item, and an evaluation unit for obtaining an evaluation value of the evaluation function 31 corresponding to the value of the parameter. The prior knowledge 34 is generated based on the plurality of evaluation values obtained by the evaluation means.

The evaluation means described above is, for example, an evaluation value map 32 that is generated by using the evaluation function 31 and that indicates the evaluation value for each value of the parameter in the evaluation item. This makes it possible to easily realize the prior knowledge generation unit 44 that generates the prior knowledge 34.

Further, the prior knowledge generation unit 44 may generate the prior knowledge 34 for each type of processing for the processing image or for each business type of the object of the processing image.

The selection unit 41 is configured to select the prior knowledge 34. More specifically, the selection unit 41 is configured to select the prior knowledge 34 based on the designated type of processing for the image or the type of business of the image object. The type of processing is, for example, inspection, positioning, character matching, or the like. Further, the business type of the target object is, for example, an automobile or food.

As described above, the prior knowledge 34 is generated for each type of processing or each business type of the object of the processing image, and the prior knowledge 34 is selected based on the designated type or business type, whereby the processing type or It is possible to obtain the prior knowledge 34 that is common to each business type of the object.

Note that the selection unit 41 may select the prior knowledge 34 based on the image given when selecting the prior knowledge 34. In this case, the selection unit 41 searches for a learning image similar to the given image using a machine-learned network or the like, and selects the learning data of the learning data set 33 including the similar learning image. Find out. Then, the selection unit 41 selects the prior knowledge 34 obtained by using the learning data.

The setting unit 42 is configured to set the value of the parameter based on the prior knowledge 34, the plurality of processing images, the evaluation items corresponding to the plurality of processing images, and the target values of the evaluation items. .. This makes it possible, for example, to search for a value in an appropriate range with an appropriate range of change for a parameter, so that an appropriate parameter value can be set in a shorter time than when there is no prior knowledge. can do. Therefore, the values of the plurality of parameters can be efficiently set.

More specifically, the setting unit 42 is configured to set the value of the parameter so that the evaluation value becomes high and the evaluation value becomes high. This makes it possible to set the value of the parameter close to or satisfying the target value of the evaluation item. Further, as described above, the evaluation function determined by the evaluation item and the target value of the evaluation item is used, and the evaluation means for obtaining the evaluation value of the evaluation function corresponding to the value of the parameter and the evaluation means are obtained. By generating the a priori knowledge 34 based on a plurality of evaluation values, it is possible to search a value in an appropriate range with an appropriate change range for a parameter suitable for increasing the evaluation value. Therefore, the value of the parameter having a higher evaluation value can be set in a shorter time than in the case without the prior knowledge 32.

The prior knowledge 34 preferably includes at least one of the degree of influence of parameters on evaluation items, the degree of dependency between parameters, the range of parameter value changes, and the range of parameter value changes. This makes it possible to easily realize the setting unit 42 that sets a parameter having a higher evaluation value in a short time.

Further, the setting unit 42 is configured to set a value having the highest evaluation value as a parameter within the specified set time range when the set time is specified. As a result, the value of the parameter having a higher evaluation value can be set even within the limited setting time.

The designation of the set time is performed, for example, by the user (user) operating the input device 70. The set time is specified from the prepared options such as 10 seconds, 1 minute, and 1 day.

The extraction unit 43 is configured to extract at least one of the plurality of parameters based on the prior knowledge 34. When the extraction unit 43 extracts the parameter, the setting unit 42 sets the value of the extracted parameter. With this, it is possible to set the value of the parameter that can increase the evaluation value among the plurality of parameters.

Each function of the control unit 40 can be realized by a program executed by a computer (microprocessor). Therefore, each function of the control unit 40 can be realized by hardware, software, or a combination of hardware and software, and is not limited to either case.

When each function of the control unit 40 is realized by software or a combination of hardware and software, the processing can be executed by multitasking, multithreading, or both multitasking and multithreading. It is not limited to such a case.

Next, the operation of the image processing apparatus according to the embodiment will be described with reference to FIG. FIG. 8 is a flowchart showing a schematic operation for setting parameters of the image processing apparatus 50 according to the embodiment.

For example, when the parameter setting is selected by the operation of the user (user), the control unit 40 of the image processing device 50 executes the parameter setting process S200 shown in FIG.

In the following description, it is assumed that the prior knowledge 34 is generated in advance and stored in the storage unit 30. That is, the prior knowledge generation step of generating the prior knowledge 34 using the plurality of learning images, the evaluation items corresponding to the plurality of learning images, and the target value of the evaluation items is performed before the parameter setting process S200. It is supposed to be done in. Further, it is assumed that a plurality of processing images, which are targets of the processing performed by the image processing device 50 based on a plurality of parameters, are given in advance, and the type of the processing or the type of business of the target is designated in advance. ..

First, the selection unit 41 takes out some of the given plurality of processing images as samples (hereinafter referred to as “sample images”) (S201).

Next, the selection unit 41 selects the prior knowledge 34 based on the designated type of processing or the type of business of the target (S202). The type of processing or the type of business of the object is specified by, for example, a user (user) operating the input device 70.

Next, the setting unit 42 uses the sample image extracted in step S201 to generate a rough evaluation value map based on the prior knowledge 34 selected in step S202 (S203). In step S203, the setting unit 42 refers to, for example, the variation width of the parameter value, the variation range of the parameter value, and the like in the prior knowledge 34.

Next, the extraction unit 43 extracts a plurality of combinations of a parameter having a high evaluation value and its value as parameter candidates based on the rough evaluation value map created in step S203 (S204).

Next, the setting unit 42 determines whether or not there is a parameter value for which the evaluation value becomes higher based on the a priori knowledge 34 selected in step S202 among the parameter candidates extracted in step S204. The surroundings are searched (S205).

As a specific example of the search, the setting unit 42 sets the value of the parameter candidate to the median value and sets the change width to the minimum value. Then, the setting unit 42 obtains the evaluation value by changing the value of the parameter within the range of the change width of the prior knowledge 34. At this time, the setting unit 42 refers to the degree of influence of the parameter on the evaluation item, the degree of dependence between the parameters, and the like in the prior knowledge 34, and determines the parameter having a high degree of influence in the evaluation item having a low evaluation value, The evaluation value is obtained by changing the value of the parameter having a high degree of dependence between the two.

Further, during step S205, the setting unit 42 writes and stores the prior knowledge 34 used in the search, the searched parameters, the search range, and the change width at the time of search in the search history 35 of the storage unit 30. Further, the setting unit 42 writes and stores the parameter, the value, the evaluation value, and the like when the evaluation value is obtained in the search, in the calculation result 36 of the storage unit 30. Thus, when the conditions such as the parameter and its value are the same, the setting unit 42 can omit the calculation of the evaluation value by referring to the calculation result 36 of the storage unit 30.

Next, the extraction unit 43 extracts a plurality of combinations of a parameter having a high evaluation value and its value as promising parameter candidates based on the search result in step S205 (S206).

Next, the setting unit 42 determines whether or not there is a parameter value for which the evaluation value becomes higher, based on the a priori knowledge 34 selected in step S202, among the promising parameter candidates extracted in step S206. Around the value is searched (S207).

As a specific example of the search, like the search in step S205, the setting unit 42 sets the value of the promising parameter candidate to the median value and sets the change width to the minimum value. Then, the setting unit 42 obtains an evaluation value by changing the parameter value in the range of the variation range of the parameter value of the prior knowledge 34. At this time, the setting unit 42 refers to, among the prior knowledge 34, the degree of influence that the parameter has on the evaluation item, the degree of dependence between the parameters, and the like, and sets the parameter having the high degree of influence in the evaluation item having the low evaluation value and the parameter. The evaluation value is obtained by changing the value of the parameter having a high degree of dependence between the two.

Also, during step S207, the setting unit 42 writes and stores the prior knowledge 34 used in the search, the searched parameters, the search range, and the change width at the time of search in the search history 35 of the storage unit 30. Further, the setting unit 42 writes and stores the parameter, the value, the evaluation value, and the like when the evaluation value is obtained in the search, in the calculation result 36 of the storage unit 30.

Next, the setting unit 42 sets the one having the highest evaluation value as the parameter value based on the result of the search in step S207 (S208).

Next, the setting unit 42 causes the output device 80 to output the contents of the search history 35 stored in the storage unit 30 via the input/output I/F 20 (S209). Thereby, the value of the set parameter has a descriptive property to the user (user), and the user (user) can estimate how much room is left for increasing the evaluation value.

After step S209, the control unit 40 ends the parameter setting process S200.

Note that when the set time is specified when executing the parameter setting process S200, the control unit 40 adjusts the value of the parameter within the range of the set time. Specifically, in step S201, the selection unit 41 extracts the number of sample images according to the designated set time. Further, in step S203, the setting unit 42 generates a rough evaluation value map and an evaluation value map with the change width of the parameter value according to the designated set time. Further, in steps S204 and S206, the extraction unit 43 extracts a number of parameter candidates and promising parameter candidates according to the designated set time.

The exemplary embodiments of the present invention have been described above. According to the image processing apparatus 50, the image processing program, and the image processing method according to the embodiment of the present invention, the prior knowledge 34 about the parameters, the plurality of processing images, the evaluation items corresponding to the plurality of processing images, and The value of the parameter is set based on the target value of the evaluation item. This makes it possible, for example, to search for a value in an appropriate range with an appropriate range of change for a parameter, so that an appropriate parameter value can be set in a shorter time than when there is no prior knowledge. can do. Therefore, the values of the plurality of parameters can be efficiently set.

The embodiments described above are for facilitating the understanding of the present invention and are not for limiting the interpretation of the present invention. Each element provided in each embodiment and the arrangement, material, condition, shape, size, and the like thereof are not limited to the exemplified ones, and can be appropriately changed. Further, the configurations shown in different embodiments can be partially replaced or combined.

(Appendix)
1. An image processing device (50) for performing processing on a processing image based on a plurality of parameters,
A prior knowledge generation unit (44) that generates prior knowledge (34) about the parameters using a plurality of learning images, evaluation items corresponding to the plurality of learning images, and target values of the evaluation items;
A priori knowledge (34), a plurality of processing images, an evaluation item corresponding to the plurality of processing images, and a setting unit (42) for setting a value of a parameter based on a target value of the evaluation item. ,
Image processing device (50).
8. An image processing program that causes a computer to perform processing on a processing image based on a plurality of parameters,
A prior knowledge generation step of generating a priori knowledge (34) about the parameter using a plurality of learning images, evaluation items corresponding to the plurality of learning images, and target values of the evaluation items;
A priori knowledge (34), a plurality of processing images, an evaluation item corresponding to the plurality of processing images, and a setting step of setting a parameter value based on a target value of the evaluation item,
Image processing program.
15. An image processing method for performing processing on a processing image based on a plurality of parameters,
A prior knowledge generation step of generating a priori knowledge (34) about the parameter using a plurality of learning images, evaluation items corresponding to the plurality of learning images, and target values of the evaluation items;
A prior knowledge (34), a plurality of processing images, an evaluation item corresponding to the plurality of processing images, and a setting step of setting a parameter value based on a target value of the evaluation item,
Image processing method.

10... Communication unit, 20... Input/output I/F, 30... Storage unit, 31... Evaluation function, 32... Evaluation value map, 33... Learning data set, 34... Prior knowledge, 35... Search history, 36... Calculation result , 40... Control unit, 41... Selection unit, 42... Setting unit, 43... Extraction unit, 44... Prior knowledge generation unit, 50... Image processing device, 51... Bus, 60... Imaging device, 70... Input device, 80... Output device, 90... Line device, 100... Image processing system, NW... Network, p1, p2, p3... Parameter, S200... Parameter setting process, SV... Server.

Claims

An image processing apparatus that performs processing on a processing image based on a plurality of parameters,
A prior knowledge generation unit that generates a priori knowledge about the parameter using a plurality of learning images, evaluation items corresponding to the plurality of learning images, and target values of the evaluation items;
A setting unit that sets the value of the parameter based on the prior knowledge, the plurality of processing images, the evaluation items corresponding to the plurality of processing images, and target values of the evaluation items;
Image processing device.
The a priori knowledge generation unit uses an evaluation function defined by the evaluation item and a target value of the evaluation item, and an evaluation unit for obtaining an evaluation value of the evaluation function corresponding to the value of the parameter and the evaluation unit. Based on the plurality of evaluation values obtained in, to generate the prior knowledge,
The setting unit sets the value of the parameter so that the evaluation value is high,
The image processing apparatus according to claim 1.
The evaluation means is an evaluation value map that is generated using the evaluation function and that indicates the evaluation value for each value of the parameter in the evaluation item.
The image processing apparatus according to claim 2.
The a priori knowledge generation unit generates the a priori knowledge for each type of the processing or for each business type of the object of the processing image,
Further comprising a selection unit for selecting the prior knowledge based on the specified type or the type of industry,
The image processing device according to claim 1.
The prior knowledge includes at least one of a degree of influence of the parameter on the evaluation item, a degree of dependence between the parameters, a change width of the value of the parameter, and a change range of the value of the parameter,
The image processing apparatus according to claim 2.
The setting unit sets a value having the highest evaluation value to the parameter within a designated set time range,
The image processing apparatus according to claim 2.
Further comprising an extraction unit that extracts at least one of the plurality of parameters based on the prior knowledge,
The setting unit sets a value of the extracted parameter,
The image processing apparatus according to claim 1.
An image processing program that causes a computer to perform processing on a processing image based on a plurality of parameters,
A prior knowledge generation step of generating a priori knowledge about the parameter using a plurality of learning images, evaluation items corresponding to the plurality of learning images, and target values of the evaluation items;
A setting step of setting a value of the parameter based on the prior knowledge, the plurality of processing images, the evaluation items corresponding to the plurality of processing images, and target values of the evaluation items,
Image processing program.
The a priori knowledge generation step uses an evaluation function defined by the evaluation item and a target value of the evaluation item, and an evaluation unit for obtaining an evaluation value of the evaluation function corresponding to the value of the parameter and the evaluation unit. Generating the a priori knowledge based on the plurality of evaluation values obtained in
The setting step includes setting the value of the parameter so that the evaluation value is high.
The image processing program according to claim 8.
The evaluation means is an evaluation value map that is generated using the evaluation function and that indicates the evaluation value for each value of the parameter in the evaluation item.
The image processing program according to claim 9.
The a priori knowledge generating step includes generating the a priori knowledge for each type of the processing or for each business type of the target object of the processing image,
Further comprising a selecting step of selecting the prior knowledge based on the specified type or the industry.
The image processing program according to any one of claims 8 to 10.
The prior knowledge includes at least one of a degree of influence of the parameter on the evaluation item, a degree of dependence between the parameters, a change width of the value of the parameter, and a change range of the value of the parameter,
The image processing program according to any one of claims 9 to 11.
The setting step includes setting a value having the highest evaluation value in the parameter within a designated set time range,
The image processing program according to any one of claims 9 to 12.
Further comprising an extracting step of extracting at least one of the plurality of parameters based on the prior knowledge,
The setting step sets a value of the extracted parameter,
The image processing program according to any one of claims 8 to 13.
An image processing method for performing processing on a processing image based on a plurality of parameters,
A prior knowledge generation step of generating a priori knowledge about the parameter using a plurality of learning images, evaluation items corresponding to the plurality of learning images, and target values of the evaluation items;
A setting step of setting a value of the parameter based on the prior knowledge, the plurality of processing images, the evaluation items corresponding to the plurality of processing images, and target values of the evaluation items,
Image processing method.