JP2010244446A - Data processor, data processing method, and program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To easily correct a converter generated by evolutionary calculation. <P>SOLUTION: This data processor includes: a generation part for generating a converter for converting applied data for learning into data which are more closer to target data for learning by combining a plurality of processing components which process input data and output the processing result as output data; a processing part for searching the actual output data of the converter by inputting the actual input data to the converter, and making it execute the processing of each processing part; and a correction part for correcting the converter according to the reception of a correction instruction for correcting the actual output data. <P>COPYRIGHT: (C)2011,JPO&INPIT

Description

  The present invention relates to a data processing device, a data processing method, and a program.

  A filter string generation method using evolutionary computation such as a genetic algorithm or genetic programming is known. In this method, operations such as crossover, mutation, and selection are repeated a plurality of times for a filter row to generate a new filter row. According to such a filter string generation method using evolutionary computation, a filter string having a complicated structure that is optimal for each case (actual application) and difficult to obtain analytically can be obtained. It can be designed with little effort.

Masayoshi Maebuchi and two others, “Study on Image Filter Design Using Genetic Algorithm” [online], Computer Utilization Education Council, [March 20, 2008 search], Internet <URL: http: //www.ciec. or.jp/event/2003/papers/pdf/E00086.pdf>

  Here, when the filter sequence generated as described above is applied to an actual application (actual application), the evolutionary calculation is restarted from the beginning for the purpose of correcting the filter sequence according to the actual application. And the calculation cost is high.

  In order to solve the above-described problem, in the first aspect of the present invention, a plurality of processing components that process input data and output processing results as output data are combined, and given input data for learning is used for learning. A generation unit that generates a converter that converts data closer to the target data, a processing unit that inputs actual input data to the converter and executes processing of each processing component, and obtains actual output data of the converter; A data processing device including a correction unit that corrects a converter in response to receiving a correction instruction for correcting actual output data.

  It should be noted that the above summary of the invention does not enumerate all the necessary features of the present invention. In addition, a sub-combination of these feature groups can also be an invention.

FIG. 1 shows a configuration of a data processing apparatus 100 according to the present embodiment. FIG. 2 shows an example of the filter array 20 having a configuration in which the filter components 22 according to the present embodiment are combined in series. FIG. 3 shows an example of the filter row 20 having a configuration in which the filter component 22 according to the present embodiment is combined in a tree structure. FIG. 4 shows an example of a genetic operation performed on the filter row 20 having a configuration in which the filter components 22 according to the present embodiment are combined in series. FIG. 5 shows an example of a crossover operation performed on the filter array 20 having a configuration in which the filter components 22 according to the present embodiment are combined in a tree structure. FIG. 6 shows an example of a mutation operation performed on the filter row 20 having a configuration in which the filter component 22 according to this embodiment is combined in a tree structure. FIG. 7 shows an example of a similarity calculation method that is an example of a parameter that represents the degree of matching of the filter array 20 according to the present embodiment. FIG. 8 shows a processing flow of the data processing apparatus 100 according to the present embodiment. FIG. 9 shows an example of the processing flow of the data processing apparatus 100 in step S840 of FIG. 2 shows an exemplary hardware configuration of a computer 1900 according to an embodiment of the present invention.

  Hereinafter, the present invention will be described through embodiments of the invention, but the following embodiments do not limit the invention according to the claims. In addition, not all the combinations of features described in the embodiments are essential for the solving means of the invention.

  FIG. 1 shows a configuration of a data processing apparatus 100 according to the present embodiment. The data processing apparatus 100 corrects a converter that converts actual input data into actual output data in accordance with a correction instruction or editing of the actual output data.

  FIG. 1 shows a configuration of a data processing apparatus 100 according to the present embodiment. The data processing apparatus 100 includes a learning data storage unit 110, a generation unit 120, a converter storage unit 130, a processing unit 140, a correction unit 150, an editing unit 160, and an adding unit 170.

  The learning data storage unit 110 stores learning data. The learning data includes learning input data and learning target data. The learning target data is data that is a target of data obtained by the converter converting the learning input data. In the present embodiment, the converter may be a filter array, the learning input data may be a learning input image, and the learning target data may be a learning target image. For example, the learning input image may be an inspection target image obtained by imaging an inspection target including a defect, and the learning target image is an extraction image of the defect that is a target to be extracted from the inspection target image. Also good.

  The generation unit 120 combines a plurality of processing components that process input data and output processing results as output data, and generates a converter that converts the given learning input data to data closer to the learning target data. . In the present embodiment, the input data may be an input image, and the output data may be an output image. In this embodiment, the processing component may be a filter component, and the converter may be a filter row including the filter component. In that case, the generation unit 120 combines a plurality of filter components that respectively convert the input image into the output image, and generates a filter row that converts the given learning input image into an image closer to the learning target image.

  In one example, the generation unit 120 includes a converter generation unit 122 and a converter selection unit 124. The converter generation unit 122 functions as a filter generation unit that performs a genetic operation described later with reference to FIGS. The converter selection unit 124 functions as a filter selection unit that calculates a fitness described later with reference to FIG. 7 and selects a filter row in accordance with the fitness.

  The converter storage unit 130 stores a plurality of different converters that have already been generated. The converter may be configured by combining a plurality of processing components that convert an input data set including one or more input data into an output data set including one or more output data. The converter may have a configuration in which processing components are combined in series. Further, the converter may have a configuration in which processing components are combined in a tree structure or the like. Further, the converter may be a program that performs an operation on the data set, an arithmetic expression that represents the operation content to be performed on the data set, or the like.

  The data set to be converted by the converter may be a one-dimensional data string, a two-dimensional data group, a three-dimensional data group, or a further multidimensional data group. The one-dimensional data string is, for example, time series data or an array data string. The two-dimensional data group is image data in which a plurality of pixel data and the like are arranged in a two-dimensional space. The three-dimensional data group is volume data or the like in which data values representing color or density are arranged at each grid point in the three-dimensional space. The converter may output a data set having a dimension different from that of the input data set.

  The processing unit 140 processes the converter when the converter is used in an actual application. Specifically, the processing unit 140 inputs actual input data to the converter, causes each of a plurality of processing components included in the converter to execute processing of the actual input data, and outputs the actual output of the converter. Ask for data. In one example, the processing unit 140 inputs an actual input image to the filter array, converts the actual input image by each of a plurality of filter components included in the filter array, and obtains an actual output image of the filter array. The actual input image may be an image obtained by imaging the surface of an inspection object that may have a defect such as a minute linear scratch. The actual output image may be an image in which a defective portion in the actual input image is emphasized.

  The correction unit 150 corrects the converter in response to receiving a correction instruction for correcting the actual output data. In one example, the correction unit 150 corrects the filter row in response to receiving a correction instruction. The correction unit 150 may correct the filter parameter of the specified filter component among the plurality of filter components included in the filter row in response to receiving the correction instruction. Here, the filter parameter of the filter component is, for example, a characteristic of the filter component such as a frequency threshold value in low-pass filtering calculation and high-pass filtering calculation, reference luminance used in an expansion filter (Dirate filter) that is one of image filters, or a filter. It may be a value that determines strength and the like.

  Further, the correction unit 150 may correct the number of repetitions in the repetition structure of two or more filter parts among the plurality of filter parts included in the filter row in response to receiving the correction instruction. In this way, when a correction instruction is received from the user after the converter is generated, the correction unit 150 corrects the converter without starting the evolutionary calculation from the beginning.

  The editing unit 160 applies a means for correcting the actual output image to the user. The adding unit 170 may have a graphical user interface for allowing the user to correct the actual output image. In this way, the user can easily correct the actual output image, for example, in order to make it easier to visually recognize the defective portion.

  The adding unit 170 stores learning data as a set of a new learning input image and a new learning target image by combining the set of the actual input image and the corrected actual output image in response to the correction of the actual output image. Add to part 110.

  Subsequently, the generation unit 120 combines a plurality of processing components, and regenerates a filter row that converts each learning input image stored in the storage unit into an image closer to the corresponding learning target image. In one example, the converter generation unit 122 has at least a part of a plurality of filter components included in the filter row included in the filter row group with respect to the filter row group including the filter row that outputs the actual output image to be corrected. A new filter row generated by replacing the filter parts with at least one other filter part is added. The converter selection unit 124 converts the original learning input image and the new learning input image into the images closer to the original learning target image and the new learning target image, respectively, among the filter columns of the filter column group. Select the filter column you want. In this way, when the filter sequence is actually applied, the generation unit 120 can generate a filter sequence that outputs an actual output image that is more preferable for the user.

  Subsequently, the processing unit 140 inputs a new actual input image to the regenerated filter sequence and causes each of a plurality of filter components included in the regenerated filter sequence to execute the process of the new actual input image. Then, an actual output image of the regenerated filter array is obtained.

  FIG. 2 shows an example of the filter row 20 having a configuration in which the filter components 22 are combined in series. FIG. 3 shows an example of the filter row 20 having a configuration in which the filter components 22 are combined in a tree structure.

  The filter array 20 receives input image data, performs a filter operation process on the received input image data, and outputs output image data. The filter row 20 may be a program that performs an operation on image data. Further, the filter array 20 may be an arithmetic expression representing the content of calculation to be performed on the image data.

  The filter row 20 has a configuration in which a plurality of filter components 22 are combined. For example, as shown in FIG. 2, the filter row 20 has a configuration in which filter components 22 are combined in series. Moreover, the filter row | line | column 20 may have the structure which combined the filter component 22 with the tree structure, as FIG. 3 shows.

  In the filter row 20 having the structure in which the filter parts 22 are combined in a tree structure, input image data is given to the filter part 22 at the end of the tree structure, and output image data is output from the uppermost filter part 22 in the tree structure. To do. Further, in such a filter row 20, the same input image data is given to each of the plurality of terminal filter components 22. Alternatively, in such a filter row 20, different input image data may be given to each of the plurality of end filter components 22.

  Each of the plurality of filter components 22 may be a program module, an arithmetic expression, or the like. The filter component 22 receives the image data output from the filter component 22 arranged in the preceding stage, performs an operation on the received image data, and gives it to the filter component 22 arranged in the subsequent stage.

  Each of the plurality of filter components 22 is a unary operation such as a binarization operation, a histogram operation, a smoothing operation, an edge detection operation, a morphology operation and / or an operation in a frequency space (for example, a low-pass filtering operation and a high-pass filtering operation). May be performed. Further, each of the plurality of filter components 22 includes an average operation, a difference operation, and / or a fuzzy operation (for example, logical sum operation, logical product operation, algebraic sum, algebraic product, limit sum, limit product, intense sum and intense product). Binary operations such as these may be performed.

  FIG. 4 shows an example of a genetic operation performed on the filter row 20 having a configuration in which the filter components 22 are combined in series. FIG. 5 shows an example of a crossover operation performed on the filter row 20 having a configuration in which the filter parts 22 are combined in a tree structure. FIG. 6 shows an example of a mutation operation performed on the filter row 20 having a configuration in which the filter parts 22 are combined in a tree structure.

  The converter generation unit 122 performs a crossover operation as an example of a genetic operation on two or more filter rows 20 to generate two or more new filter rows 20. As shown in FIGS. 4 and 5, the converter generation unit 122 converts a part of the filter component group 24 </ b> A of at least one filter row 20 </ b> A that has already been generated into another filter row 20 </ b> B that has already been generated. New filter rows 20E and 20F may be generated by replacing at least a part of the filter component group 24B. The filter component group 24 is a member in which one or a plurality of filter components 22 are combined.

  The converter generation unit 122 performs a mutation operation as an example of a genetic operation on one filter row 20 to generate a new filter row 20. As shown in FIGS. 4 and 6, the converter generation unit 122 replaces a part of the filter component group 24 </ b> C of the already generated filter row 20 </ b> C with another filter component group 24 </ b> G selected at random. Then, a new filter row 20G may be generated.

  Further, the converter generation unit 122 may leave the current generation filter row 20 as it is as the next generation filter row 20. As illustrated in FIG. 4, the converter generation unit 122 may generate a next-generation filter row 20H that includes the configuration of the filter component 22 of the filter row 20D as it is.

  The converter selection unit 124 selects one or a plurality of filter rows 20 by a technique in which a natural selection of living organisms is modeled with respect to the plurality of filter rows 20 generated by the converter generation unit 122. The converter selection unit 124 preferentially selects a filter row 20 having a higher degree of matching among the plurality of filter rows 20. The converter selection unit 124 may select the filter row 20 according to a technique such as elite selection and roulette selection based on the degree of matching of each of the plurality of filter rows 20. Then, the converter selection unit 124 stores the filter row 20 in the converter storage unit 130 so that the selected filter row 20 can survive to the next generation, and stores the filter row 20 that has not been selected in the converter memory. It is deleted from the part 130.

  FIG. 7 shows an example of a method for calculating similarity, which is an example of a parameter representing the degree of matching of the filter row 20. In the present embodiment, the converter selection unit 124 calculates a similarity indicating how similar the output image generated by converting the learning input image by the filter row 20 and the target image are. . This similarity is used as an evaluation value or index indicating that the larger the value is, the more appropriate the filter array 20 that generated the output image is.

  The converter selection unit 124 calculates a comparison value by comparing the value of each region of the output image with the value of each region corresponding to the target image, and sets the weight specified by the weight data for the comparison value for each region. Multiply. Then, the converter selection unit 124 calculates a value obtained by averaging or summing the comparison values for each region multiplied by the weight for all regions, and outputs the calculated value as the similarity of the output image.

  In one example, the converter selection unit 124 calculates the difference or ratio between the luminance value for each pixel of the output image and the luminance value of the corresponding pixel of the target image. Then, the converter selection unit 124 multiplies each calculated difference or ratio for each pixel by the corresponding weight specified by the weight data, and sums or averages the difference or ratio for each pixel multiplied by the weight. The similarity may be calculated.

  For example, when the weight data is represented by a weight image, the converter selection unit 124 may calculate the similarity by performing an operation represented by the following formula (1). In this case, the sizes of the output image, the target image, and the weight image are the same.

  In the formula (1), f represents the similarity. Ymax represents the maximum luminance value.

  X is a variable indicating the pixel position in the horizontal direction of the image. y is a variable indicating the pixel position in the vertical direction of the image. xmax is a constant indicating the number of pixels in the horizontal direction of the image. ymax is a constant indicating the number of pixels in the vertical direction of the image.

  Iweight (x, y) represents the luminance value of the pixel at the (x, y) position in the weighted image. Itarget (x, y) represents the luminance value of the pixel at the (x, y) position in the target image. Ifilter (x, y) represents the luminance value of the pixel at the (x, y) position in the output image.

  That is, as shown in the numerator part in the curly braces of Expression (1), the converter selection unit 124 determines the luminance value of the weighted image with respect to the absolute value of the difference between the luminance value of the target image and the luminance value of the output image. Is calculated for every pixel in the screen, and a total weighted difference value is calculated by adding the calculated weighted difference values for all pixels. Further, as shown in the denominator part in the curly braces of Equation (1), the converter selection unit 124 calculates a total weight obtained by adding the luminance values of the weighted image for all pixels. Furthermore, the converter selection unit 124 normalizes the division value obtained by dividing the total weighting difference value by the total weight (in the curly braces of the expression (1)) by the reciprocal (1 / Ymax) of the maximum value of the luminance value. The value (the second term in equation (1)) is calculated. Then, the converter selection unit 124 calculates a value obtained by subtracting the normalized value from 1 as the similarity f. In this way, the converter selection unit 124 can calculate the similarity by weighting the difference between the output image and the target image for each region.

  FIG. 8 shows a processing flow of the data processing apparatus 100. The data processing apparatus 100 repeatedly executes the processes in steps S810 to S860 multiple times (for example, multiple generations) (S810, S860). Note that the data processing apparatus 100 uses a filter string 20 that is randomly generated or has already been generated as an initial filter string 20 for generating the filter string 20 that converts the given input data for learning into learning target data. , Stored in the converter storage unit 130 in advance.

  In each generation, first, the converter generation unit 122 performs a genetic operation such as a crossover operation and a mutation operation on the plurality of filter arrays 20 remaining from the previous generation, and thereby creates a new plurality of filter arrays 20. Is generated (S820). In the first generation, the converter generation unit 122 may perform a genetic operation on the plurality of filter arrays 20 generated in advance by a user or the like to generate a plurality of new filter arrays 20. Good.

  Subsequently, the converter selection unit 124 filters the learning input image with each of the plurality of filter arrays 20 remaining from the previous generation and the plurality of filter arrays 20 newly generated in step S820, and outputs an output image. Generate (S830). Thereby, the converter selection unit 124 generates a plurality of output images corresponding to the plurality of filter rows 20.

  Subsequently, the converter selection unit 124 calculates the similarity between each of the plurality of output images and the learning target image, and selects the plurality of filter rows 20 corresponding to the output images with higher similarity (S840). . Note that the converter selection unit 124 may select one filter row 20 corresponding to one output image having the highest similarity in the last generation.

  Subsequently, the converter selection unit 124 causes the one or more filter arrays 20 selected in step S840 to remain in the next generation, and deletes the filter array 20 that generated the output image not selected in step S840 (S850). ). The data processing apparatus 100 repeatedly executes the above processing for a plurality of generations (for example, several tens or hundreds of generations), executes the processing up to the Nth generation (N is an integer of 2 or more), and then exits the flow. (S860). In this way, the data processing apparatus 100 can generate the filter array 20 suitable for converting the learning input image into the learning target image using the evolutionary calculation.

  Subsequently, the processing unit 140 performs data conversion in an actual application using the filter array 20 generated in steps S810 to S860. The processing unit 140 obtains an actual output image by converting an actual input image such as a captured image obtained by imaging the inspection object (S870). The editing unit 160 receives a correction instruction for correcting the filter row 20 generated in steps S810 to S860 from the user (S880).

  If there is no correction instruction in step S880, the process returns to S870, and data conversion in the actual application is continued.

  When there is a correction instruction in step S880, the correction unit 150 corrects the filter parameter of the filter component 22 included in the filter row 20 according to the correction instruction (S890). Subsequently, the correction unit 150 corrects the number of repetitions in the repetition structure of the filter component 22 included in the filter row 20 in accordance with the correction instruction (S892). The data processing apparatus 100 may modify the filter parameter and / or the number of repetitions of the filter component in accordance with designation from the user.

  If there is a correction of the actual output image, the correction unit 150 displays the actual output image obtained in step S870 and corrects the image by the user (S894, S896). If the actual output image is not corrected, the data processing apparatus 100 returns the process to S870 and continues the data conversion in the actual application.

  Subsequently, the data processing apparatus 100 repeats the processes of Steps S810 to S860 a plurality of times again using the actual input image and the corrected actual output image as the learning input image and the learning target image, respectively. In that case, the generation unit 120 may use the current gene (eg, the current filter row 20) as the initial gene.

  Here, the step of correcting the filter parameter, the step of correcting the number of repetitions of the filter component 22, the step of correcting the image by the user, and the step of repeating the genetic process on the filter row 20 are all performed. Need not be executed, and may be executed selectively and in any order.

  FIG. 9 shows an example of the processing flow of the data processing apparatus 100 in step S840 of FIG. The data processing apparatus 100 executes the following processing in step S840 shown in FIG.

  First, the converter selection unit 124 executes the processing of the following step S920 for each of the plurality of output images generated by each of the plurality of filter arrays 20 stored in the converter storage unit 130 (S910, S930). In step S920, the converter selection unit 124 compares the output image with the learning target image by weighting according to the weighted image, and calculates the similarity between the output image and the learning target image. When the converter selection unit 124 finishes the processing for all of the plurality of output images, the converter proceeds to step S940 (S930).

  Subsequently, the converter selection unit 124 selects an output image closer to the learning target image from the plurality of output images based on the respective similarities of the plurality of output images (S940). In one example, the converter selection unit 124 preferentially selects an output image having a higher degree of similarity among a plurality of output images converted by the plurality of filter arrays 20. The converter selection unit 124 may select an output image whose similarity is higher than the reference similarity. Moreover, the converter selection part 124 may select the output image of the range from which the similarity degree was predetermined from the upper rank. Moreover, the converter selection part 124 may select an output image at random on the conditions set so that an output image with higher similarity may be selected with a higher probability.

  Subsequently, the converter selection unit 124 selects the filter sequence 20 that generates the output image selected by the converter selection unit 124 as the filter sequence 20 that converts the learning input image into a similar image by the learning target image. (S950). When the process of step S950 is completed, the data processing apparatus 100 ends the flow.

  FIG. 10 shows an example of a hardware configuration of a computer 1900 according to the embodiment of the present invention. A computer 1900 according to this embodiment is connected to a CPU peripheral unit having a CPU 2000, a RAM 2020, a graphic controller 2075, and a display device 2080 that are connected to each other by a host controller 2082, and to the host controller 2082 by an input / output controller 2084. An input / output unit having a communication interface 2030, a hard disk drive 2040, and a DVD drive 2060, and a legacy input / output unit having a ROM 2010, a flexible disk drive 2050, and an input / output chip 2070 connected to the input / output controller 2084. Prepare.

  The host controller 2082 connects the RAM 2020 to the CPU 2000 and the graphic controller 2075 that access the RAM 2020 at a high transfer rate. The CPU 2000 operates based on programs stored in the ROM 2010 and the RAM 2020 and controls each unit. The graphic controller 2075 acquires image data generated by the CPU 2000 or the like on a frame buffer provided in the RAM 2020 and displays it on the display device 2080. Instead of this, the graphic controller 2075 may include a frame buffer for storing image data generated by the CPU 2000 or the like.

  The input / output controller 2084 connects the host controller 2082 to the communication interface 2030, the hard disk drive 2040, and the DVD drive 2060, which are relatively high-speed input / output devices. The communication interface 2030 communicates with other devices via a network. The hard disk drive 2040 stores programs and data used by the CPU 2000 in the computer 1900. The DVD drive 2060 reads a program or data from the DVD 2095 and provides it to the hard disk drive 2040 via the RAM 2020.

  The input / output controller 2084 is connected to the ROM 2010, the flexible disk drive 2050, and the relatively low-speed input / output device of the input / output chip 2070. The ROM 2010 stores a boot program that the computer 1900 executes at startup and / or a program that depends on the hardware of the computer 1900. The flexible disk drive 2050 reads a program or data from the flexible disk 2090 and provides it to the hard disk drive 2040 via the RAM 2020. The input / output chip 2070 connects the flexible disk drive 2050 to the input / output controller 2084 and inputs / outputs various input / output devices via, for example, a parallel port, a serial port, a keyboard port, a mouse port, and the like. Connect to controller 2084.

  A program provided to the hard disk drive 2040 via the RAM 2020 is stored in a recording medium such as the flexible disk 2090, the DVD 2095, or an IC card and provided by the user. The program is read from the recording medium, installed in the hard disk drive 2040 in the computer 1900 via the RAM 2020, and executed by the CPU 2000.

  A program that is installed in the computer 1900 and causes the computer 1900 to function as the data processing apparatus 100 according to the first embodiment includes a learning data storage module, a generation module, a converter storage module, a processing module, a correction module, an editing module, and an addition Module. These programs or modules work on the CPU 2000 or the like to change the computer 1900 into the learning data storage unit 110, the generation unit 120, the converter storage unit 130, the processing unit 140, the correction unit 150, the editing unit 160, and the addition unit 170. As each function.

  The information processing described in these programs is read by the computer 1900, whereby the learning data storage unit 110, the generation unit 120, which are specific means in which the software and the various hardware resources described above cooperate. It functions as a converter storage unit 130, a processing unit 140, a correction unit 150, an editing unit 160, and an adding unit 170. And the specific data processing apparatus 100 according to the intended use is constructed | assembled by implement | achieving the calculation or processing of the information according to the intended use of the computer 1900 in 1st Embodiment by these specific means.

  In an example, when communication is performed between the computer 1900 and an external device or the like, the CPU 2000 executes a communication program loaded on the RAM 2020, and based on the processing contents described in the communication program, the communication interface A communication process is instructed to 2030. Under the control of the CPU 2000, the communication interface 2030 reads transmission data stored in a transmission buffer area or the like provided on a storage device such as the RAM 2020, the hard disk drive 2040, the flexible disk 2090, or the DVD 2095, and transmits it to the network. Alternatively, the reception data received from the network is written into a reception buffer area or the like provided on the storage device. As described above, the communication interface 2030 may transfer transmission / reception data to / from the storage device by a DMA (direct memory access) method. Instead, the CPU 2000 transfers the storage device or the communication interface 2030 as a transfer source. The transmission / reception data may be transferred by reading the data from the data and writing the data to the communication interface 2030 or the storage device of the transfer destination.

  In addition, the CPU 2000 DMAs all or necessary portions from among files or databases stored in an external storage device such as the hard disk drive 2040, the DVD drive 2060 (DVD 2095), and the flexible disk drive 2050 (flexible disk 2090). The data is read into the RAM 2020 by transfer or the like, and various processes are performed on the data on the RAM 2020. Then, CPU 2000 writes the processed data back to the external storage device by DMA transfer or the like. In such processing, since the RAM 2020 can be regarded as temporarily holding the contents of the external storage device, in the present embodiment, the RAM 2020 and the external storage device are collectively referred to as a memory, a storage unit, or a storage device. Various types of information such as various programs, data, tables, and databases in the present embodiment are stored on such a storage device and are subjected to information processing. Note that the CPU 2000 can also store a part of the RAM 2020 in the cache memory and perform reading and writing on the cache memory. Even in such a form, the cache memory bears a part of the function of the RAM 2020. Therefore, in the present embodiment, the cache memory is also included in the RAM 2020, the memory, and / or the storage device unless otherwise indicated. To do.

  In addition, the CPU 2000 performs various operations, such as various operations, information processing, condition determination, information search / replacement, etc., described in the present embodiment, specified for the data read from the RAM 2020 by the instruction sequence of the program. Is written back to the RAM 2020. For example, when performing the condition determination, the CPU 2000 determines whether the various variables shown in the present embodiment satisfy the conditions such as large, small, above, below, equal, etc., compared to other variables or constants. When the condition is satisfied (or not satisfied), the program branches to a different instruction sequence or calls a subroutine.

  Further, the CPU 2000 can search for information stored in a file or database in the storage device. For example, in the case where a plurality of entries in which the attribute value of the second attribute is associated with the attribute value of the first attribute are stored in the storage device, the CPU 2000 displays the plurality of entries stored in the storage device. The entry that matches the condition in which the attribute value of the first attribute is specified is retrieved, and the attribute value of the second attribute that is stored in the entry is read, thereby associating with the first attribute that satisfies the predetermined condition The attribute value of the specified second attribute can be obtained.

  The program or module shown above may be stored in an external recording medium. As the recording medium, in addition to the flexible disk 2090 and the DVD 2095, an optical recording medium such as a CD, a magneto-optical recording medium such as an MO, a tape medium, a semiconductor memory such as an IC card, and the like can be used. Further, a storage device such as a hard disk or RAM provided in a server system connected to a dedicated communication network or the Internet may be used as a recording medium, and the program may be provided to the computer 1900 via the network.

  As mentioned above, although this invention was demonstrated using embodiment, the technical scope of this invention is not limited to the range as described in the said embodiment. It will be apparent to those skilled in the art that various modifications or improvements can be added to the above-described embodiment. It is apparent from the scope of the claims that the embodiments added with such changes or improvements can be included in the technical scope of the present invention.

  The order of execution of each process such as operations, procedures, steps, and stages in the apparatus, system, program, and method shown in the claims, the description, and the drawings is particularly “before” or “prior to”. It should be noted that the output can be realized in any order unless the output of the previous process is used in the subsequent process. Regarding the operation flow in the claims, the description, and the drawings, even if it is described using “first”, “next”, etc. for convenience, it means that it is essential to carry out in this order. It is not a thing.

  20 filter rows, 22 filter parts, 24 filter parts group, 100 data processing device, 110 learning data storage unit, 120 generation unit, 122 converter generation unit, 124 converter selection unit, 130 converter storage unit, 140 processing unit , 150 correction unit, 160 editing unit, 170 addition unit, 1900 computer, 2000 CPU, 2010 ROM, 2020 RAM, 2030 communication interface, 2040 hard disk drive, 2050 flexible disk drive, 2060 DVD drive, 2070 input / output chip, 2075 graphics Controller, 2080 display device, 2082 Host controller, 2084 I / O controller, 2090 flexible disk, 2095 DVD

Claims (8)

A generator that generates a converter that processes input data and outputs a processing result as output data in combination with a plurality of processing components and converts the given learning input data into data closer to the learning target data;
A processor that inputs actual input data to the converter and causes each of the plurality of processing components included in the converter to execute processing to obtain actual output data of the converter;
A correction unit for correcting the converter in response to receiving a correction instruction for correcting the actual output data;
A data processing apparatus comprising: The generation unit combines a plurality of filter components that respectively convert an input image into an output image, and generates a filter row that converts a given learning input image into an image closer to the learning target image,
The processing unit inputs an actual input image to the filter row, converts each of the plurality of filter components included in the filter row, and obtains an actual output image of the filter row,
The data processing apparatus according to claim 1, wherein the correction unit corrects the filter row in response to receiving the correction instruction.   The data processing apparatus according to claim 2, wherein the correction unit corrects a filter parameter of a specified filter component among the plurality of filter components included in the filter row in response to receiving the correction instruction.   4. The correction unit according to claim 2, wherein the correction unit corrects the number of repetitions in a repetition structure of two or more filter parts among the plurality of filter parts included in the filter row in response to receiving the correction instruction. Data processing equipment. A storage unit for storing a set of the learning input image and the learning target image;
In response to the correction of the actual output image, the set of the actual input image and the corrected actual output image is stored in the storage unit as a set of the new learning input image and the new learning target image. It further includes an additional part to be added,
The generating unit combines the plurality of processing components to regenerate the filter sequence for converting each learning input image stored in the storage unit into an image closer to the corresponding learning target image. ,
The processing unit inputs a new actual input image to the regenerated filter row and causes each of the plurality of filter components included in the regenerated filter row to execute the regenerated The data processing apparatus according to any one of claims 2 to 4, wherein an actual output image of the filter array is obtained. The generator is
At least a part of the plurality of filter components included in the filter row included in the filter row group with respect to the filter row group including the filter row that has output the actual output image to be corrected. A filter generation unit for adding a new filter row generated by replacing with one other filter component;
The original learning input image and the new learning input image are converted into images closer to the original learning target image and the new learning target image, respectively, among the filter rows of the filter row group. A filter selection unit for selecting the filter row;
The data processing apparatus according to claim 5, comprising: A step of generating a converter that combines a plurality of processing components that process input data and output processing results as output data, and converts the given learning input data to data closer to the learning target data by the generation unit When,
A step of inputting actual input data to the converter by the processing unit, causing each of the plurality of processing components included in the converter to be executed, and obtaining actual output data of the converter;
Correcting the converter in response to receiving a correction instruction for correcting the actual output data by a correction unit;
A data processing method comprising: A program for causing a computer to function as a data processing device,
The computer,
A generator that generates a converter that processes input data and outputs a processing result as output data in combination with a plurality of processing components and converts the given learning input data into data closer to the learning target data;
A processor that inputs actual input data to the converter and causes each of the plurality of processing components included in the converter to execute processing to obtain actual output data of the converter;
A correction unit for correcting the converter in response to receiving a correction instruction for correcting the actual output data;
Program to make it work.

Images