JP2007241945A - Image processor and image processing determination device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an image processor for arithmetically processing a numerical index designating an adaptability or an extraction degree of extraction image at high speed in an evolutionary image processing method for extracting an image corresponding to a target image from an original image by performing processing by three images of the original image, the target image and a weighted image and two or more kinds of preliminarily prepared image processing filters. <P>SOLUTION: An extraction image group storage part 37 stores extraction images resulted from processing of the original image stored in an original image storage part 33 with filters stored in an image processing filter group storage part 32 in the order of filter trees of a filter tree storage part 36, and an adaptability storage part 39 stores an adaptability calculated based on the finally extracted extraction image, the target image and the weighted image by a fitness calculation part 38, and outputs it from a control part 31. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention provides an evolutionary image processing method for extracting an image corresponding to a target image from an original image by performing processing using three images of an original image, a target image, and a weighted image, and a plurality of types of image processing filters prepared in advance. In (Non-Patent Document 1), the present invention relates to an image processing apparatus for performing high-speed arithmetic processing on fitness (Non-Patent Document 1, equation (5.3) on page 97). In addition, in the evolutionary image processing method, the image processing determination device is configured to make a high-speed determination using a numerical index designating the degree of extraction of an extracted image.

  In the conventional evolutionary image processing method (Non-Patent Document 1, page 87), a plurality of filters in an image processing filter group are combined in a computer using three images of an original image, a target image, and a weighted image as input information. A filter (tree filter) is generated, a plurality of filter processes are performed on the original image according to the filter tree, and an extracted image processed according to the filter tree from the original image as an index of the appropriateness of the filter tree; Up to the calculation of the fitness with the target image has been processed by the program in the computer, but the amount of processing is enormous, and there is a problem that it takes a lot of time.

In the evolutionary image processing method, a filter (tree filter) is generated by combining a plurality of filters in the image processing filter group in the computer using the three images of the original image, the target image, and the weighted image as input information. Then, a plurality of filter processes are performed on the original image according to the filter tree, and the fitness of the extracted image processed according to the filter tree from the original image and the target image is calculated as an index of the appropriateness of the filter tree. However, the amount of processing is enormous and it takes a lot of time.
Tomoharu Nagao “Evolutionary Image Processing” Shoshodo, May 2002, p. 1-192

  In the evolutionary image processing method, input information specifying the original image, target image, weighted image, and image processing filter group and a filter (tree filter) that combines multiple filters from the filter group. Thus, the fitness is calculated at high speed from the image extracted from the original image, the target image, and the weighted image.

  Also, in the evolutionary image processing method, the original image, the image processing filter group and a filter (tree filter) that combines a plurality of filters from the filter group, and the extraction condition of the image that is the result of filtering the original image are determined. By inputting a specified determination processing method and a determination value, an extraction degree of an image extracted from an original image is determined at high speed.

An image processing apparatus according to the present invention includes:
A control unit that performs input / output control of required information;
An original image storage unit for storing an original image input from the control unit;
A target image storage unit for storing a target image input from the control unit;
A weight image storage unit for storing a weight image input from the control unit;
An image processing filter group storage unit for storing image processing filter groups;
A filter tree storage unit for storing a filter tree input from the control unit;
A fitness calculation unit for storing a calculation program of fitness;
A processing sequence control unit that controls processing by the image processing filter and a fitness calculation sequence based on information of the filter tree stored in the filter tree storage unit in accordance with an instruction from the control unit;
The information of the filter tree stored in the filter tree storage unit is read by the control of the processing sequence control unit, and the original image stored in the original image storage unit is processed in the order of the filter tree by the filter stored in the image processing filter group storage unit An extracted image group storage unit for storing the extracted image as a result of applying
The extracted image last extracted by the control of the processing sequence control unit, the target image of the target image storage unit, and the weight image stored in the weight image storage unit are read out and stored in the fitness calculation unit based on these images. It has an fitness storage unit that can calculate fitness by an fitness calculation program, store the result, and pass the value of the fitness to the control unit in accordance with a reading instruction from the control unit.

An image processing determination apparatus according to the present invention
A control unit that performs input / output control of required information;
An original image storage unit for storing an original image input from the control unit;
An image processing filter group storage unit for storing image processing filter groups;
A filter tree storage unit for storing a filter tree input from the control unit;
A determination processing method storage unit that stores programs according to a plurality of processing methods for determining the degree of extraction of the extracted image;
A determination processing instruction information storage unit for instructing a determination processing program of the determination processing method storage unit;
A determination value storage unit for storing a determination value for performing the determination;
Stores the determination value obtained by processing by the determination processing program of the determination processing method storage unit instructed by the determination processing instruction information storage unit, and stores the result of comparing the determination value with the determination value of the determination value storage unit A determination result storage unit to perform,
According to an instruction from the control unit, the processing unit includes a processing sequence control unit that controls a sequence such as processing by the image processing filter and determination processing based on information of the filter tree stored in the filter tree storage unit.

  In this evolutionary image processing method, information specifying the three images of the original image, the target image, and the weighted image, and a filter (tree filter) combining the image processing filter group and the filter is input from the computer to the apparatus. Since the fitness can be calculated at high speed from the image extracted from the original image, the target image, and the weighted image by, for example, in the case of realizing image processing for detecting a flaw, a flaw inspection system capable of high-speed processing can be constructed. It becomes possible.

  This method is an evolutionary image processing method that uses an original image, a filter (tree filter) that combines the image processing filter group and the filter, a determination processing program, and a determination process specified by inputting a determination value from a computer to the device. Since it is possible to calculate the degree of extraction of an image extracted by filtering from an original image with a program at high speed, for example, in the case of realizing image processing for detecting a flaw, it is possible to construct a flaw inspection system capable of high-speed processing. It becomes possible.

Embodiment 1 FIG.
Reference numerals 11, 12, and 13 in FIG. 1 are examples of an original image, a target image, and a weight image that are information that needs to be input in an image processing apparatus to which an evolutionary image processing method is applied. The evolutionary image processing method attempts to extract an image corresponding to the image indicated by the target image 12 from the original image 11.

  FIG. 2 shows an example of a filter tree, 21 is an original image to be filtered, and f1, f2, f5, f7, f10, f12 are binarization, thinning, density conversion, etc., for example. Filter instruction information for instructing the image processing filter. Reference numerals 22, 23, 24, 25, 26, 27, and 28 are extracted images sequentially extracted from the original image according to the filter instruction information of f1, f3, f10, f7, f5, and f12. An image processing filter group 29 includes a plurality of filter processing procedures.

  FIG. 3 shows an embodiment of an image processing apparatus. Reference numeral 31 denotes a control unit that performs input / output control of required information. Reference numeral 32 denotes an image processing filter that is input from the control unit 31 and includes binarization, thinning, density conversion, and the like. An image processing filter group storage unit that performs image processing in accordance with filter instruction information that is stored and input groups, 33 is an original image storage unit that stores an original image input from the control unit 31, and 34 is input from the control unit 31. A target image storage unit for storing the target image, 35 a weight image storage unit for storing a weight image input from the control unit 31, and 36 for storing a filter tree composed of filter instruction information input by the control of the control unit 31. The filter tree storage unit 37 stores the extracted image that has been subjected to filter processing in the order of the filter instruction information stored in the filter tree storage unit 36 in the original image in the original image storage unit 33. The output image group storage unit 38 is an adaptation used in evolutionary image processing from the extracted image extracted by the final image processing filter of the filter tree, the target image of the target image storage unit 34, and the weight image of the weight image storage unit 35. An fitness calculation unit 39 for calculating the degree, an fitness storage unit 39 for storing the value of the result calculated by the fitness calculation unit 38, and a filter 40 stored in the filter tree storage unit 36 under the control of the control unit 31 It is a processing sequence control unit that controls a processing sequence such as calculating fitness by processing in the order of instruction information.

  FIG. 4 is a diagram showing the configuration of the extracted image group storage unit 37 of FIG.

  First, the control unit 31 instructs the processing sequence control unit 40 to send information about the image processing filter group, the original image, the target image, the filter tree, and the fitness calculation program to the image processing filter group storage unit 32, the original image, respectively. The data is stored in the storage unit 33, target image storage unit 34, weight image storage unit 35, filter tree storage unit 36, and fitness calculation unit 37.

  Subsequently, when the control unit 31 instructs the processing sequence control unit 40 to start processing, the processing sequence control unit 40 first stores the first filter instruction information f1 in the filter tree of FIG. The original image stored in the unit 33 is read out and the image processing filter group storage unit 32 is instructed to perform the filter process.

  The image processing filter group storage unit 32 selects an image processing filter f1P (FIG. 2) that matches the filter instruction information f1 and performs filter processing on the original image input from the original image storage unit 33. The resulting extracted image is stored in the extracted image group storage unit 371 by the sequence control unit 40 (FIG. 4). Subsequently, the processing sequence control unit 40 reads out the next filter instruction information f3 and the extracted image 22 stored in the extracted image group storage unit 371 from the filter tree storage unit 36, and filters the image processing filter group storage unit 32 using f3P. Instruct processing.

  The image processing filter group storage unit 32 selects f3P of the image processing filter that matches the filter instruction information f3, and performs filter processing on the extracted image input from the extracted image group storage unit 37 (371 in FIG. 4). The resulting extracted image is stored in the extracted image group storage unit 372 (FIG. 4) by the sequence control unit 40.

  Such processing is sequentially performed according to the filter instruction information f10, f7 and f5, f12, and f13, and the filter processing of f10P, f7P, f5P, f12P, and f13P is executed, and the respective extracted images are extracted as 373 of the extracted image group storage unit 37. 374, 375, 376, and 377.

  The processing sequence control unit 40, which has detected that the f13P filter processing has been completed by the last filter instruction information f13 instructed by the filter tree storage unit 36, continues to the fitness calculation unit 38 and 377 of the extracted image group storage unit 37. The last extracted image (FIG. 4), the target image in the target image storage unit 34, and the weight image in the weight image storage unit 35 are input to instruct the start of arithmetic processing. In response to the start instruction, the fitness calculation unit 38 calculates the fitness from the extracted image, the target image, and the weighted image, and stores the resultant value in the fitness storage unit 39.

  Subsequently, when the control unit 31 instructs the processing sequence control unit 40 to read the fitness value, the processing sequence control unit 40 instructs the fitness storage unit 39 to read the fitness value, and the fitness storage unit 39 that has received the instruction. Passes the stored fitness value to the control unit 31.

  When the control unit 31 instructs the processing sequence control unit 40 to read the information 37i (i = 1, 2,...) Indicating the storage unit to the extracted image group storage unit 37, the processing sequence control unit 40 An extracted image corresponding to the instruction information in the extracted image group storage unit 37 is read and passed to the control unit 31. The above is the description of the operation.

  In the above embodiment, the description has been made assuming that each of the filter tree, the input image, the target image, and the weight image is one. However, as shown in FIG. 5, a plurality of filter trees, input images, target images, and weight images can be stored. Thus, by preparing a plurality of extracted image group storage units (see FIG. 6), fitness calculation units, and fitness storage units, a plurality of fitness values can be obtained at the same time by the same operation as described above.

  In the embodiment, the capacity of the extraction storage unit and the capacity of the image processing filter group storage unit have been described without mentioning them, but the capacity of the extraction storage unit becomes a capacity according to the number of filter instruction information, and the image processing filter Since the capacity of the group storage unit also depends on the capacity of the procedure of the processing filter to be prepared, the maximum value of the number of filter instruction information and the number of processing filters is determined in advance to determine the capacity, and the capacity of the memory installed in the apparatus is Just decide.

  The instruction to read the fitness value from the control unit 31 to the processing sequence control unit 40 has been described by the look-in method. When the calculation result is stored in the fitness storage unit 39, the control unit 31 is interrupted. If an interrupt line is prepared so that it can be performed, the interrupt method can be realized and the processing method can be speeded up.

  Similarly, the readout of the extracted image has been described by the look-in method. However, as in the case of the calculation result, if the extracted image is stored in the extracted image group storage unit 37, the interrupt method can be realized and the response can be realized. High speed.

  In this method, by inputting information specifying three images of an original image, a target image, and a weight image, and a filter (tree filter) that combines the image processing filter group and the filter into a device connected to the computer, A device independent of the computer can extract a desired image from the original image, and the fitness can be calculated from the extracted image, the target image, and the weighted image, so that the speed of evolutionary image processing can be achieved.

Embodiment 2. FIG.
In this image processing determination apparatus, the original image 11 of FIG. 1 is input.

  The filter tree is the same as in FIG.

  FIG. 7 shows an embodiment of the image processing determination apparatus. In the same manner as described above, reference numeral 31 denotes a control unit that performs input / output control of required information, 32 denotes binarization, thinning, density conversion, and the like input from the control unit 31. An image processing filter group storage unit that stores the image processing filter group and performs image processing according to the filter instruction information, 36 is a filter tree storage unit that stores a filter tree made up of filter instruction information input under the control of the control unit 31, 37 is an extracted image group storage unit that stores an extracted image obtained by filtering the original image in the original image storage unit 34 according to the order of the filter instruction information stored in the filter tree storage unit 35, and 40 is controlled by the control unit 31. The processing sequence control unit controls a series of processing sequences such as processing in the order of the filter instruction information received and stored in the filter tree storage unit 36. 71 is an evaluation processing method storage unit for storing an evaluation processing program such as counting the area ratio of the extracted image input from the control unit 31 and the length of the outer ring, and 33 is an original image input from the control unit 31. An original image storage unit for storing, a determination processing instruction information storage for storing information for specifying a determination processing program input from the control unit 31, and a determination value storage unit for storing determination values input from the control unit 31 , 74 stores the determination value calculated by the determination processing program stored in the determination processing method storage unit in accordance with the determination processing instruction information in the extracted image extracted by the final image processing filter of the filter tree and converts the value into the determination value The determination result storage unit stores the determination value stored in the storage unit and the comparison result.

  The configuration of the extracted image group storage unit 37 is the same as that of FIG.

  The operation is first instructed from the control unit 31 to the processing sequence control unit 40 to store each information of the image processing filter group, the determination processing program, the original image, the filter tree, the determination value, and the determination processing instruction information, respectively. 32, the determination processing method storage unit 71, the original image storage unit 33, the filter tree storage unit 36, the determination processing instruction information storage unit 72, and the determination value calculation unit 73.

  Subsequently, when the control unit 31 instructs the processing sequence control unit 40 to start processing, the processing sequence control unit 40 first stores the first filter instruction information f1 in the filter tree of FIG. The original image stored in the unit 33 is read out and the image processing filter group storage unit 32 is instructed to perform the filter process.

  The image processing filter group storage unit 32 selects an image processing filter f1P (FIG. 2) that matches the filter instruction information f1 and performs filter processing on the original image input from the original image storage unit 34. The resulting extracted image is stored in the extracted image group storage unit 371 by the sequence control unit 40 (FIG. 4). Subsequently, the processing sequence control unit 40 reads out the next filter instruction information f3 and the extracted image 22 stored in the extracted image group storage unit 371 from the filter tree storage unit 35, and filters the image processing filter group storage unit 32 using f3P. Instruct processing.

  The image processing filter group storage unit 32 selects f3P of the image processing filter that matches the filter instruction information f3, and performs filter processing on the extracted image input from the extracted image group storage unit 37 (371 in FIG. 4). The resulting extracted image is stored in the extracted image group storage unit 372 (FIG. 4) by the sequence control unit 40.

  Such processing is sequentially performed according to the filter instruction information f10, f7 and f5, f12, and f13, and the filter processing of f10P, f7P, f5P, f12P, and f13P is executed, and the respective extracted images are extracted as 373 of the extracted image group storage unit 37. 374, 375, 376, and 377.

  The processing sequence control unit 40 that has detected that the f13P filter processing by the last filter instruction information f13 instructed by the filter tree storage unit 36 has been completed is subsequently designated by the determination processing information of the determination processing instruction information unit 72. According to the determination processing program of the determination processing method storage unit 71, the start of the determination process of the last extracted image of 377 (FIG. 4) of the extracted image group storage unit 37 is instructed. Upon receiving the start instruction, the determination processing method storage unit 71 calculates a determination value from the extracted image, and passes the result value to the determination result storage unit 74 as a result of comparing the value with the value of the determination value storage unit 73. . The determination result storage unit 74 that has been calculated and passed the determination value and the determination result stores them.

  Subsequently, when the control unit 31 instructs the processing sequence control unit 40 to read the determination value and the determination result, the processing sequence control unit 40 instructs the determination result storage unit 74 to read the determination result and receives the instruction. 74 passes the stored determination value and determination result to the control unit 31.

  When the control unit 31 instructs the processing sequence control unit 40 to read the information 37i (i = 1, 2,...) Indicating the storage unit to the extracted image group storage unit 37, the processing sequence control unit 40 An extracted image corresponding to the instruction information in the extracted image group storage unit 37 is read and passed to the control unit 31. The above is the description of the operation.

  In the above embodiment, the original image, the filter tree, the judgment value instruction information, and the judgment value have been described as one. However, as shown in FIG. 9, a plurality of original images, filter trees, judgment value instruction information, and judgment values are provided. By preparing a plurality of extracted image group storage units and determination result storage units in a similar manner, it is possible to obtain a plurality of determination results and determination values at once by the same operation as described above.

  In the embodiment, the capacity of the extraction storage unit and the capacity of the image processing filter group storage unit have been described without mentioning them, but the capacity of the extraction storage unit becomes a capacity according to the number of filter instruction information, and the image processing filter Since the capacity of the group storage unit also depends on the capacity of the procedure of the processing filter to be prepared, the maximum value of the number of filter instruction information and the number of processing filters is determined in advance to determine the capacity, and the capacity of the memory installed in the apparatus is Just decide.

  Although the instruction to read the determination result from the control unit 31 toward the processing sequence control unit 40 has been described by the look-in method, the control unit 31 can be interrupted when the determination result is stored in the determination result storage unit 74. Thus, if an interrupt line is prepared, the interrupt method can be realized and the processing method can be speeded up.

  Similarly, the read-out of the extracted image has been described by the look-in method. However, as in the case of the determination result, the interrupt method can be realized by using a method of interrupting when the extracted image is stored in the extracted image group storage unit 38. High speed.

  This method is specified by inputting an original image, an image processing filter group and a filter (tree filter) combining the filters, a determination processing program, and a determination value from the computer to the apparatus connected to the computer. Since it becomes possible to calculate the extraction degree of the image extracted by filtering from the original image with the judgment processing program at a high speed, a desired image can be extracted from the original image on the device independent of the computer, and the extracted image Therefore, the image processing speed can be increased.

  The image processing device and the image processing determination device are computers, and each element can execute processing by a program. Further, the program can be stored in a storage medium so that the computer can read the program from the storage medium.

It is a figure which shows the example of the original image which is the information which needs to be input in the image processing apparatus to which an evolutionary image processing system is applied, a target image, and a weight image. It is a figure which shows the example of a filter tree. It is a figure which shows the Example of an image processing apparatus. 4 is a diagram illustrating a configuration of an extracted image group storage unit 37. FIG. It is a figure which shows the Example of the image processing apparatus which memorize | stores multiple filter trees, an input image, a target image, and a weight image. It is a figure which shows the extraction image group memory | storage part in the case of inputting several images. 6 is a diagram illustrating an example of an image processing device according to Embodiment 2. FIG. 3 is a diagram illustrating a configuration of a determination processing method storage unit 71. FIG. It is a figure which shows the Example of the image processing apparatus which memorize | stores several filter trees, input images, a target image, and a weight image concerning Embodiment 2. FIG.

Explanation of symbols

Original image 11, target image 12, weight image 13, image processing filter group 29, 31 control unit, 32 image processing filter group storage unit, 33 original image storage unit, 34 target image storage unit, 35 weight image storage unit, 36 filter Tree storage unit, 37 extracted image group storage unit, 38 fitness calculation unit, 39 fitness storage unit, 40 processing sequence control unit, 71 determination processing method storage unit, 72 determination processing instruction information storage unit, 73 determination value storage unit, 74 Determination result storage unit.

Claims (2)

A control unit that performs input / output control of required information;
An original image storage unit for storing an original image input from the control unit;
A target image storage unit for storing a target image input from the control unit;
A weight image storage unit for storing a weight image input from the control unit;
An image processing filter group storage unit for storing image processing filter groups;
A filter tree storage unit for storing a filter tree input from the control unit;
A fitness calculation unit for storing a calculation program of fitness;
A processing sequence control unit that controls processing by the image processing filter and a fitness calculation sequence based on information of the filter tree stored in the filter tree storage unit in accordance with an instruction from the control unit;
The information of the filter tree stored in the filter tree storage unit is read by the control of the processing sequence control unit, and the original image stored in the original image storage unit is processed in the order of the filter tree by the filter stored in the image processing filter group storage unit An extracted image group storage unit for storing the extracted image as a result of applying
The extracted image last extracted by the control of the processing sequence control unit, the target image of the target image storage unit, and the weight image stored in the weight image storage unit are read out and stored in the fitness calculation unit based on these images. An image processing system comprising an fitness storage unit capable of computing fitness by an fitness calculation program, storing the result, and passing the fitness value to the control unit in accordance with a reading instruction from the control unit. Science device. A control unit that performs input / output control of required information;
An original image storage unit for storing an original image input from the control unit;
An image processing filter group storage unit for storing image processing filter groups;
A filter tree storage unit for storing a filter tree input from the control unit;
A determination processing method storage unit that stores programs according to a plurality of processing methods for determining the degree of extraction of the extracted image;
A determination processing instruction information storage unit for instructing a determination processing program of the determination processing method storage unit;
A determination value storage unit for storing a determination value for performing the determination;
Stores the determination value obtained by processing by the determination processing program of the determination processing method storage unit instructed by the determination processing instruction information storage unit, and stores the result of comparing the determination value with the determination value of the determination value storage unit A determination result storage unit to perform,
An image processing characterized by having a processing sequence control unit that controls a sequence such as processing by an image processing filter and determination processing based on information of a filter tree stored in a filter tree storage unit in accordance with an instruction from the control unit Judgment device.

Images