JP2021051617A - Image processing system and image processing program - Google Patents

Abstract

To provide a system for optimizing image processing in image processing technology which detects an object from an image.SOLUTION: An image processing system includes: an individual storage unit which stores individual information for defining the order of executing function units in a network having a closed circuit; an image storage unit which stores a learning image and a verification image; a teacher storage unit which stores teacher information associated with the learning image and the verification image and discriminating a detection area for evaluating output information from a non-detection area; an image processing unit which processes the learning image and the verification image on the basis of the individual information to generate output information; an evaluation value calculation unit which compares the output information with the teacher information and calculates an evaluation value of the individual information to rank the individual information; and a gene manipulation unit which selects or modifies the individual information on the basis of the rank information of the individual information to update the individual storage unit. The evaluation value calculation unit calculates an evaluation value of the detection area of the learning image and the verification image and an evaluation value of the non-detection area in different calculation methods for the detection area and the non-detection area of the teacher information, to calculate an evaluation value of individual information.SELECTED DRAWING: Figure 1

Description

The present invention relates to an image processing technique for finding a detection target from an image, and more particularly to an image processing system for optimizing image processing.

Conventionally, in image processing technology for finding a detection target from an image, the image processing can be optimized by comparing the output image obtained by performing image processing on the input image with the teacher image and calculating the evaluation value. It has been.
In this optimization method, the evaluation value may deteriorate based on the ratio of the detected region to the undetected region in the image, it takes time to create a teacher image, and a person due to computer optimization. The problem was that excessive interference was required.

Japanese Patent No. 5479944 Japanese Patent No. 4862150 Japanese Patent No. 501153 Japanese Unexamined Patent Publication No. 2008-15817

Here, as an image processing technique for finding a detection target from an image, the method described in Patent Document 1 can be mentioned.
In this method, size-dependent crossover is introduced into a parallel image filter automatic generation system by genetic programming (GP). As a result, by selecting and adopting an image with cracks from multiple actual pavement images as training data for filter construction, an image filter for extracting cracks is automatically constructed from various types of images, and this is used. Damage levels are being assessed.

Further, as a technique using such so-called evolutionary computation, the evolutionary computation system described in Patent Documents 2 to 4 can be mentioned. According to these evolutionary computation systems, it is possible to optimize the processing algorithm in each system.

However, even if these techniques are used, when the image processing is optimized by comparing the output image obtained by performing image processing on the input image with the teacher image, as described above, the detection target region There are problems that the evaluation value deteriorates depending on the ratio, that it takes time and effort to generate a teacher image, and that excessive human interference is required.
For example, if the output image is evaluated by the difference in brightness for each pixel using the least squares method in order to compare it with the teacher image, the evaluation value is based on the size of the detection target even if the same non-detection is performed. There was a big difference in.

The present inventors consider that such an influence on the evaluation value of the size of the detection target is an obstacle to the optimization of image processing, and consider that the range of the detection region (the region where the detection target exists). It was conceived that the above problem can be solved by creating the information indicating the above as teacher information and comparing the output information obtained by performing image processing on the input image with this teacher information.
In addition, it is possible to improve the detection performance by using different methods suitable for each of the detection area and the non-detection area (the area where the detection target does not exist).
Furthermore, the present invention has been completed by making it possible to improve the degree of freedom of optimization by adopting the method of genetic network programming (GNP) in evolutionary computation.

That is, an object of the present invention is to provide an image processing system and an image processing program capable of further optimizing image processing in an image processing technique for finding a detection target from an image.

In order to achieve the above object, the image processing system of the present invention optimizes image processing for detecting a detection target from an image by using genetic manipulation, which is an optimization method that mathematically simulates the evolution of living organisms. A processing system that stores a population storage unit that stores multiple individual information that defines the execution order of multiple functional units in a network that can include closed paths, and stores multiple learning images and multiple verification images. An image storage unit to be used, a teacher information storage unit that stores teacher information that separates a detection area and a non-detection area for evaluating output information, which is associated with the learning image and the verification image, respectively, and the above-mentioned. An image processing unit that processes the learning image and the verification image based on the individual information and generates the output information, compares the output information with the teacher information, and calculates an evaluation value of the individual information. The evaluation value calculation unit that ranks the individual information based on the evaluation value calculated using the learning image, and the individual group is selected or changed based on the ranking information of the individual information. The evaluation value calculation unit includes a genetic manipulation unit that updates the storage unit, and the evaluation value calculation unit uses different calculation methods for the detection region and the non-detection region of the teacher information to obtain the detection region of the learning image and the verification image. The evaluation value of the above and the evaluation value of the non-detection region are calculated, and the evaluation value of the individual information is calculated based on these evaluation values.

Further, it is preferable that the image processing system of the present invention has a configuration in which the evaluation value of the detection region is calculated by point distribution analysis.
Further, it is preferable that the image processing system of the present invention has a configuration in which the evaluation value of the non-detection region is calculated based on the difference in brightness from the teacher information.

Further, the image processing program of the present invention is an image processing program that optimizes image processing for detecting a detection target from an image by using a genetic operation which is an optimization method that mathematically simulates the evolution of an organism. , A computer, a population storage unit that stores a plurality of individual information that defines an execution order of a plurality of functional units in a network that may include a closed path, an image storage unit that stores a learning image and a verification image, the learning A teacher information storage unit that stores teacher information that separates a detection area and a non-detection area for evaluating output information, which is associated with the image for verification and the image for verification, and processes the image based on the individual information. Then, the image processing unit that generates the output information calculates the evaluation value by comparing the output information with the teacher information, and ranks the individual information based on the evaluation value calculated using the learning image. The evaluation value calculation unit and the evaluation value calculation unit are used to select or change the individual information based on the ranking information of the individual information to function as a gene manipulation unit for updating the population storage unit. The evaluation value of the detection area and the evaluation value of the non-detection area of the learning image and the verification image are calculated by different calculation methods for the detection area and the non-detection area of the teacher information, and these evaluation values are used. Originally, the structure is such that the evaluation value of the individual information is calculated.

According to the present invention, it is possible to provide an image processing system and an image processing program capable of further optimizing image processing in an image processing technique for finding a detection target from an image.

It is a block diagram which shows the structure of the image processing apparatus of 1st Embodiment of this invention. It is a flowchart which shows the processing procedure by the image processing apparatus of 1st Embodiment of this invention. It is a flowchart which shows the processing procedure of a gene manipulation in the processing procedure by the image processing apparatus of 1st Embodiment of this invention. It is a block diagram which shows the structure of the image processing apparatus of the 2nd Embodiment of this invention. It is a flowchart which shows the processing procedure by the image processing apparatus of the 2nd Embodiment of this invention. It is explanatory drawing which shows the test condition of a reference example and an Example. It is explanatory drawing which shows the function (image processing: noise removal) of a node used in a reference example and an Example. It is explanatory drawing which shows the function (image processing: luminance correction) of a node used in a reference example and an Example. It is explanatory drawing which shows the function (image processing: binarization) of the node used in the reference example and the Example. It is explanatory drawing which shows the function (image processing: edge detection) of a node used in a reference example and an Example. It is explanatory drawing which shows the function (image processing: frequency filter) of a node used in a reference example and an Example. It is explanatory drawing which shows the function (image processing: calculation) of a node used in a reference example and an Example. It is explanatory drawing which shows the function (image processing: morphology) of a node used in a reference example and an Example. It is explanatory drawing which shows the function (image processing: detection target determination) of the node used in a reference example and an Example. It is explanatory drawing which shows the function (image processing: branch) of a node used in a reference example and an Example. It is explanatory drawing which shows the step of making the image for learning / verification and teacher information in a reference example and an Example. It is a figure which shows the graph which shows the transition of the normalized evaluation value in a reference example and an Example. It is a figure which shows the structure of the individual information obtained by learning of the reference example, and the function of the node used in the individual information. It is a figure which shows the structure of the individual information obtained by learning of an Example, and the function of the node used in the individual information. It is a figure which shows the sensitivity and peculiarity of the learning result and the verification result by a reference example. It is a figure which shows the sensitivity / specificity of the learning result and the verification result by an Example. It is a figure which shows the example of the correct answer image and incorrect answer image by a reference example and an Example.

Hereinafter, an image processing system of the present invention and an embodiment of an image processing program will be described in detail. However, the present invention is not limited to the specific contents of the following embodiments and examples described later.

[First Embodiment]
First, the image processing system according to the first embodiment of the present invention and the first embodiment of the image processing program will be described with reference to FIGS. 1 to 3. FIG. 1 is a block diagram showing a configuration of an image processing device corresponding to the image processing system of the present embodiment, and FIG. 2 is a flowchart showing a processing procedure by the image processing device. Further, FIG. 3 is a flowchart showing a processing procedure of gene manipulation in the processing procedure by the image processing apparatus.

The image processing system of the present embodiment optimizes image processing for detecting a detection target from an image by using genetic manipulation, which is an optimization method that mathematically simulates the evolution of living organisms.
Specifically, as shown in FIG. 1, the image processing device 1 of the present embodiment has a functional unit storage unit 10, an individual generation unit 11, a population storage unit 12, an image input unit 13, an image storage unit 14, and a teacher. It includes an information input unit 15, a teacher information storage unit 16, an image processing unit 17, an output information storage unit 18, an evaluation value calculation unit 19, an evaluation result storage unit 20, and a gene manipulation unit 21.
As shown in FIG. 1, all of these configurations in the image processing device 1 of the present embodiment can be provided in one information processing device. Further, each of these configurations may be distributed and provided in each device of the image processing system including a plurality of information processing devices. This also applies to the second embodiment described later.

The functional unit storage unit 10 stores nodes (corresponding to genes), which are functional units that execute a certain process, for each node number. In the present embodiment, various nodes for performing image processing can be used as this node, and in the examples described later, those shown in FIGS. 6 to 14 are used.
Further, when a parameter used for executing each process exists, these nodes are stored in the functional unit storage unit 10 together with the parameter.

Further, as the node, a node that does not execute any processing may be used, or a node (node set) that is a collection of functional units that execute a plurality of processes may be used. By using the nodes including those that do not perform any processing, the actual number of nodes can be reduced, for example, when the set value of the initial number of nodes for generating individual information is too large.

In the present specification, "image processing" includes various filter processing for noise removal, correction processing such as luminance correction, binarization processing, edge detection processing, frequency filter processing, arithmetic processing such as four-rule calculation, and morphology. Processing, detection target determination processing, branching processing, etc. are included.

The individual generation unit 11 generates individual information composed of an array of a plurality of nodes by using the nodes stored in the functional unit storage unit 10. At this time, the individual generation unit 11 can randomly generate a plurality of individual information based on the plurality of nodes.
At this time, the individual generation unit 11 can define the execution order of a plurality of nodes in a network shape that can include a closed cycle in the individual information. This network is a directed network consisting of oriented links and may contain feedback. The network can also include linear and tree structures.

By using such individual information, it is possible to perform learning including a network structure having more flexible expressive power as compared with the case of using individual information consisting only of a linear structure or a tree structure. It has an advantage in deriving the optimum solution for practical use and the practical solution for practical use). That is, it is difficult for a person to know in advance which structure is the optimum solution for a problem, so it is difficult to determine the conditions of the structure before machine learning. Therefore, searching for the optimal solution under a wide range of conditions using the individual information that defines the execution order of multiple nodes in a network that can include a cycle may lead to the true optimal solution. It is considered to be effective in enhancing.

The population storage unit 12 stores a population composed of a plurality of individual information generated by the individual generation unit 11. As this population, one generation or a plurality of generations can be stored in the population storage unit 12, and individual information can be stored in the population storage unit 12 for each generation number.
The evaluation value calculation unit 19, which will be described later, can rank individual information for each generation and store the ranking information in the evaluation result storage unit 20.

The image input unit 13 inputs a plurality of learning images and a plurality of verification images to the image processing device 1, and stores each image identification information in the image storage unit 14.
The teacher information input unit 15 inputs teacher information for evaluating the output information, and stores the corresponding image identification information in the teacher information storage unit 16.

In the present embodiment, the teacher information is associated with the learning image and the verification image, respectively, and is information for dividing the detection area and the non-detection area for evaluating the output information, and may be referred to as a teacher range. ..
As this teacher information, for example, in the learning image and the verification image, the part where the detection target exists is surrounded by a specific color (white in the examples described later), and the part where the detection target does not exist is another color. By showing (black in the examples described later), it is possible to obtain an image that separates the detection area and the non-detection area.
The shape of the detection region surrounding the portion where the detection target exists is preferably one in which teacher information can be easily created, and is preferably a rectangle, a circle, an ellipse, or the like.

Here, in the present embodiment, a defect that is some kind of aggregate is assumed as a detection target. Such defects include, for example, wrinkles in the film, splashes of ink, and scratches such as those rubbed with sandpaper.
In the case of such an aggregate, there is a problem that it takes a very long time to create a teacher image as teacher information. Further, the present inventors have already developed a method of using the number of teachers and the teacher position as teacher information, but there is a problem that it is difficult to specify the number and position when the detection target is an aggregate.
Therefore, the present inventors have made it possible to further reduce the time and effort required to create a teacher image by using the teacher range as the teacher information.

Further, in the present embodiment, in the creation of the teacher information (teacher range) of the learning image and the verification image, the teacher information (teacher range) is created without setting the detection area for the target that should not be detected. Is preferable.
There are generally the following three types of objects that should not be detected.
(1) Unevenness, pattern, holes for bleeding air from the mold, etc. that exist as a design, etc. (2) Even if it is a defect, scratches (small, thin scratches, etc.) that fit within the specifications, inspection Dust that can be excluded in the subsequent process (3) As a result of the film etc. coming close, there is no problem in itself, but halation is reflected in the image

When the teacher information is used as an image, the evaluation value is strongly affected by the size of the detection target in the evaluation by the least squares method, so the shape of the large detection target is different from that when there is no small detection target. However, the evaluation value tends to deteriorate.
For example, suppose that when a triangle is detected in a certain image, a triangle having a size of 10 pixels is not detected.
At this time, if the output of the larger triangle in the same image is one size smaller than that of the teacher image, if the difference is 10 pixels or more, the evaluation value will be worse than the absence of the 10 pixel triangle. Become. For this reason, in machine learning, the evaluation value tends to be better when a triangle larger than a small triangle is brought closer to the teacher image, evolution progresses so as to match a triangle larger than a small triangle, and it becomes easy to fall into a local solution. ..
On the other hand, in the present embodiment, for images having different sizes of detection targets, the error in the case of no detection is the same, so that it is difficult to fall into a local solution.

The image processing unit 17 inputs an image from the image storage unit 14, and also inputs individual information from the population storage unit 12, and sequentially executes a plurality of nodes included in the individual information based on the individual information. The image processing unit 17 also executes a node for executing each of the image input processing and the output image storage processing.

At this time, each node executes the process corresponding to each function by using the information obtained from the image. Further, each node can selectively execute the next node corresponding to the processing result based on the processing result. Therefore, even if the nodes are included in the individual information, not all the nodes are necessarily executed in the image processing. Of course, all the nodes in the individual information may be executed. Further, by giving feedback, a node defined as only one in one individual information may be executed a plurality of times.

In addition, as described above, the node functions include various filter processing for noise removal, correction processing such as luminance correction, binarization processing, edge detection processing, frequency filter processing, and arithmetic processing such as four-rule calculation. There are image processing in a narrow sense such as morphology processing and branch processing. These are mainly performed by the image processing unit 17 as a process for finding a detection target candidate.

Further, in the detection target determination process as a function of the node, the image processing unit 17 labels the detection target candidates in a certain range in the image, calculates the feature amount, and provides the feature amount based on the threshold value or the like. It is performed as a process of determining whether or not the detection target candidate is a correct detection target.
In the detection target determination process, it is also possible to perform a process of labeling a non-detection target candidate, calculating its feature amount, and determining that it is not a detection target because the feature amount is not provided based on a threshold value or the like. ..

Further, the image processing unit 17 can create an image or the like as output information for each of the learning image and the verification image and store the image in the output information storage unit 18.
At this time, for the learning image and the verification image, if the detection target is detected, an image with information indicating the area where the detection target exists is added, and if the detection target is not detected, the information is added. An image that is not added is created.
Then, this output information can be stored in the output information storage unit 18 for each image identification information of the image and for each identification information of individual information.

When a detection target exists in the learning image and the verification image, the evaluation value calculation unit 19 can be made to perform a function of generating an image or the like to which information indicating the range is added. That is, after image processing is performed by the image processing unit 17, the evaluation value calculation unit 19 generates an image to which information indicating the range of the detection target is added, stores the image in the output information storage unit 18, and uses this to store the evaluation value. It is also possible to have a configuration that calculates.

The evaluation value calculation unit 19 inputs output information from the output information storage unit 18, and also inputs teacher information corresponding to the image identification information corresponding to the output information from the teacher information storage unit 16.
Then, the evaluation value is calculated by comparing the output information and the teacher information, and the obtained evaluation value is stored in the evaluation result storage unit 20 for each image identification information of the image and for each identification information of the individual information. it can.

At this time, the evaluation value calculation unit 19 calculates the evaluation value of the detection area and the evaluation value of the non-detection area of the learning image and the verification image by different calculation methods for the detection area and the non-detection area of the teacher information. In addition, the evaluation value of the individual information is calculated based on the evaluation value calculated using the learning image. Specifically, the evaluation value of the individual information can be calculated by adding up the evaluation value of the detection area and the evaluation value of the non-detection area.
At this time, it is preferable that the evaluation value calculation unit 19 calculates the evaluation value of the detection region by point distribution analysis.

Here, in the present embodiment, it is necessary to grasp the quality of the detection status in the detection region by evolutionary computation, but when the detection target is some kind of aggregate, the index that reacts on average in the detection region is Since it was considered good, point distribution analysis was adopted.
In the present embodiment, it is preferable to use at least one of the K-function method, the nearest neighbor distance, and the partition method as the point distribution analysis.

At this time, it is preferable that the evaluation value calculation unit 19 calculates the evaluation value of the non-detection region based on the brightness difference between the non-detection region and the teacher information.
Here, it is premised that there is no detection target in the non-detection area. For example, when noise or an object that is a non-detection target but is characteristic is shown in the non-detection area, these May be falsely detected. Therefore, in order to evaluate that the detection target is not detected in the non-detection area, the evaluation value is calculated for the non-detection area.
Note that "not detected" means a case where the brightness difference between the output information and the teacher information in the non-detection region is 0 (least squares error = 0). That is, if 0 <the least squares error of the non-detection region, it is known that something has been detected, and if it is 0, it is not detected.

Ａ：非検出領域の面積
ｙ_Ｍ(i)：検出領域を分割した各領域において検出有りとして反応したピクセル数
Ｍ：分割数
ａ：全検出領域内で検出有りとして反応したピクセル数
ｗ：重み係数
なお、検出領域において何も反応がない場合（検出されない場合）、期待値μは0となり数式では発散するが、この場合はプログラム上で発散を抑制して評価値を意図的に悪い値にすることなどにより処理することが可能である。 Specifically, the evaluation value calculation by the evaluation value calculation unit 19 can be preferably performed by using, for example, the following formula.

A: Area of non-detection area y _M (i): Number of pixels that reacted as detected in each area that divided the detection area M: Number of divisions a: Number of pixels that reacted as detected in all detection areas w: Weight coefficient If there is no reaction in the detection region (when it is not detected), the expected value μ becomes 0 and diverges in the mathematical formula, but in this case, the divergence is suppressed on the program and the evaluation value is intentionally made a bad value. It is possible to process by such things.

Further, when the evaluation value calculation unit 19 finishes calculating the evaluation value for all the individual information (or the individual information for one generation) stored in the population storage unit 12 and stores it in the evaluation result storage unit 20. For example, individual information can be ranked (ranked) based on these evaluation values, and the obtained ranking information can be stored in the evaluation result storage unit 20.

Further, the evaluation value calculation unit 19 can perform an end determination for determining whether or not to end the image processing by the image processing unit 17. The end determination can be performed based on whether or not the end condition is satisfied, and if the end condition is satisfied, the image processing by the image processing unit 17 can be terminated and the processing by the image processing device 1 can be terminated. Then, the individual information having the highest evaluation value stored in the evaluation result storage unit 20 is obtained as the optimized individual information for performing image processing.

Examples of the termination condition include the case where image processing is completed for a preset population of all generations, and the case where evolution has not occurred in a certain number of generations or more.
The case where evolution has not occurred is the case where the evaluation value is not higher than the time point of comparison.
The end determination function may be separated from the evaluation value calculation unit 19, and the image processing device 1 of the present embodiment may include an end determination unit for performing the end determination.

The gene manipulation unit 21 updates the population storage unit 12 by selecting or changing the individual information based on the ranking information of the individual information in the evaluation result storage unit 20.
Specifically, the genetic manipulation unit 21 includes an elite individual storage unit that leaves individual information with a high evaluation value to the next generation, and an individual selection unit that leaves individual information to the next generation with a certain probability based on the evaluation value. It can have a crossing part that exchanges a part of two individual information with each other, and a mutation part that randomly rewrites a part or all of one selected individual information or a node parameter.

The elite individual preservation unit can unconditionally leave the individual information with the best evaluation value in the population of each generation to the next generation.
The individual selection unit can select individual information having a high evaluation value in the population of each generation with a high probability and leave it to the next generation.

As a crossing process, the crossing portion can perform one-way crossing in which the same (node number) arrangement in the array of each node of a plurality of individual information is replaced with a certain probability. In addition, as a crossing process, the crossing part is a one-point crossing in which one point in the same sequence is used as a boundary in the array of each node and one of the sequences is exchanged with each other, or an array is mutually bounded by a plurality of points in the same row. It is also possible to perform multi-point crossing to be replaced. The crossing section can perform such crossing processing on the individual information selected by the individual selection section.

As a mutation process, the mutation unit rewrites a part of the individual information or rewrites all of the individual information with a certain probability for each individual information using the node stored in the functional unit storage unit 10. be able to. The mutation unit can perform such mutation processing on the individual information selected by the individual selection unit. In addition, the mutation unit can also generate new individual information by using the node stored in the functional unit storage unit 10 as the mutation process. Further, it is also possible to cause the mutation portion to perform parameter mutation as a mutation process by randomly rewriting a part or all of the parameters of the node of the selected parent individual information.

If the number of individuals of the same series generated based on the same individual information increases, diversity will be lost and evolution will enter a dead end (a state in which it has fallen into a mathematically local solution). On the other hand, if the number of individuals other than the same series is increased too much, it approaches a random search and efficient evolution is not performed.
Therefore, in order to have diversity in the early stage of learning, processing by the mutation part is performed so as to increase the number of individuals other than the same series, and in the middle stage of learning, the same series is processed by the individual selection part in order to carry out efficient evolution. It is preferable to increase it. Further, when the evolution is stopped, it is possible to preferably increase the number of mutations other than the same series by the mutation site in order to give diversity again.

Then, the gene manipulation unit 21 generates a new generation population and stores it in the population storage unit 12. At this time, the genetic manipulation unit 21 can update the population storage unit 12 by overwriting the population of the previous generation. In addition, the population storage unit 12 can be updated by storing a population of a plurality of generations for each generation.

Here, the learning image and the verification image used in the present embodiment will be described in detail. In this embodiment, a holdout method is used in which the image for learning and the image for verification are divided into two.
In the present embodiment, the learning image is processed by all individual information (image processing algorithm) to obtain output information, and the output information is compared with the teacher information to calculate the evaluation value, and the evaluation value. Can be used to perform individual ranking using.
In addition, the verification image is processed by the best individual information (image processing algorithm) to obtain output information, and the evaluation value is calculated by comparing the output information with the teacher information, and the evaluation value. Can be used to verify the validity of the analysis model and whether over-learning has occurred.
Since the verification image is used by a person to analyze the result, it basically does not affect the evolutionary computation, but when the evaluation value of the verification image is used for the end judgment, for example, the verification image Evolutionary computation will be affected if it ends when the evaluation value reaches the threshold, or if it ends early when there is a large difference between the evaluation values of the learning image and the verification image.

When the verification image is processed with the best individual information to obtain the output information, it can be performed for each generation and the best individual information in each generation can be used. It can also be carried out for each set generation (for example, every 10 generations) with the best individual information. Furthermore, it can be performed at the timing when the evaluation value using the learning image is improved or at the finally obtained individual information (after the evolutionary calculation is completed: the final generation). In this way, the timing for processing the verification image can be arbitrarily set.
In the examples described later, when the evaluation value using the learning image is improved, the corresponding individual information is processed to obtain the output information of the verification image.

Next, the processing procedure by the image processing apparatus of this embodiment will be described with reference to FIGS. 2 and 3.
First, the image input unit 13 in the image processing device 1 inputs an image (step 10). Further, the teacher information input unit 15 inputs a teacher range as teacher information (step 11). As the teacher range, for example, an image showing the division between the detection area and the non-detection area can be used. Then, the individual generation unit 11 generates an initial population (step 12).

At this time, the individual generation unit 11 may generate individual information by randomly selecting and arranging nodes from the functional unit storage unit 10, and store each identification information of the individual information in the population storage unit 12. it can. In addition, the individual generation unit 11 can store a population of a plurality of generations for each generation number in the population storage unit 12 with a population consisting of a plurality of individual information as one generation.

Next, the image processing unit 17 executes image processing for all the individual information in the population for each generation of the population in the population storage unit 12 (steps 13, 14, 15).
At this time, the image processing unit 17 can create an image or the like as output information for each of the learning image and the verification image, and store the image in the output information storage unit 18. At this time, when the detection target exists in the learning image and the verification image, the image processing unit 17 creates an image to which the information indicating the region where the detection target exists is added and stores it in the output information storage unit 18. Can be made to. Although the processing timings of the learning image and the verification image are different as described above, they are described together for convenience.

Next, the evaluation value calculation unit 19 compares the output information with the teacher information for each individual information and calculates the evaluation value (step 16).
At this time, in the calculation of the evaluation value by the evaluation value calculation unit 19, the evaluation value of the detection region is calculated by the point distribution analysis such as the partition method. In addition, the evaluation value of the non-detection region is calculated based on the brightness difference between the non-detection region and the teacher information. Then, the evaluation value of the detection area and the evaluation value of the non-detection area are added up to calculate the evaluation value of the individual information.
Further, the evaluation value calculation unit 19 stores the obtained evaluation value in the evaluation result storage unit 20 for each individual information.
Then, for all the individual information of the relevant generation, the processing from the image processing to the calculation and storage of the evaluation value is repeatedly executed.

Next, the evaluation value calculation unit 19 ranks the individual information based on the evaluation value stored in the evaluation result storage unit 20 (step 17).
In addition, the evaluation value calculation unit 19 performs a determination process of whether or not the end condition is satisfied (step 18). If the end condition is satisfied, the process by the image processing device 1 is terminated.

If the termination condition is not satisfied, the gene manipulation unit 21 executes the gene manipulation process (step 19).
Specifically, as shown in FIG. 3, the elite individual preservation unit in the gene manipulation unit 21 selects the individual information having the highest evaluation value as the individual information of the next-generation population (step 20).
Further, the individual selection unit in the gene manipulation unit 21 selects as individual information of the next-generation population with a higher probability as the individual information having a higher evaluation value ranks (step 21).

Further, the crossing portion in the gene manipulation unit 21 executes a crossing process on the individual information selected by the individual selection unit to generate new individual information (step 22).
In addition, the mutation unit in the gene manipulation unit 21 performs mutation processing or the like based on the individual information selected by the individual selection unit or without using these individual information to generate new individual information ( Step 23).
In addition, the order of each process by the elite individual preservation unit, the individual selection unit, the cross section, and the mutation unit may be exchanged within the range of the order in which each process can be executed.

According to the image processing apparatus of the present embodiment, in the image processing technique for finding the detection target from the image, the evaluation value deteriorates depending on the ratio of the detection target area by using the information indicating the range of the detection area as the teacher information. Can solve the problem. In addition, it is possible to solve the problem that it has been troublesome to generate a teacher image and the problem that excessive human interference has been required in the past. Further, it is possible to improve the detection performance by using different methods suitable for each of the detection region and the non-detection region.

[Second Embodiment]
Next, the image processing system and the image processing program according to the second embodiment of the present invention will be described with reference to FIGS. 4 and 5. FIG. 4 is a block diagram showing a configuration of an image processing device corresponding to the image processing system of the present embodiment, and FIG. 5 is a flowchart showing a processing procedure by the image processing device.

As shown in FIG. 4, the image processing apparatus 1a of the present embodiment includes a functional unit storage unit 10a, an individual generation unit 11a, a population storage unit 12a, an image input unit 13a, an image storage unit 14a, and a teacher information input unit 15a. It includes a teacher information storage unit 16a, an image processing unit 17a, an output information storage unit 18a, an evaluation value calculation unit 19a, an evaluation result storage unit 20a, and a gene manipulation unit 21a. Further, the population storage unit 12a includes a parent population storage unit 121a and a child population storage unit 122a.

The functional unit storage unit 10a, the image input unit 13a, the image storage unit 14a, the teacher information input unit 15a, the teacher information storage unit 16a, the output information storage unit 18a, and the evaluation result storage unit 20a in the image processing device 1a of the present embodiment are , The same as in the first embodiment. Further, other configurations in the image processing apparatus 1a of the present embodiment can be the same as those of the first embodiment except for the points described below.

The individual generation unit 11a generates individual information composed of an array of a plurality of nodes by using the nodes stored in the functional unit storage unit 10a. At this time, the individual generation unit 11a can define the execution order of a plurality of nodes in a network shape that can include a closed cycle in the individual information, as in the first embodiment.
Further, the individual generation unit 11a in the present embodiment randomly generates a plurality of parent individual information as individual information, and stores an initial parent population composed of these parent individual information in the parent population storage unit 121a. it can.

The gene manipulation unit 21a executes a gene manipulation on the parent individual information in the parent population storage unit 121a.
Specifically, first, parent individual information for generating child individual information is randomly selected from the parent population in the parent population storage unit 121a.

The elite individual preservation unit in the genetic manipulation unit 21a can store a child population consisting of the same child individual information as the selected parent individual information in the child population storage unit 122a.
Then, the crossing portion in the genetic manipulation unit 21a executes a crossing process on the selected parent individual information to generate a plurality of child individual information, and stores the child population composed of these child individual information as the child population. It can be stored in the unit 122a.
Further, the mutation unit in the genetic manipulation unit 21a executes mutation processing on the selected parent individual information using the node stored in the functional unit storage unit 10a, and the parent individual has a certain probability. A child population consisting of child individual information obtained by rewriting a part of the information or rewriting all of the information can be stored in the child population storage unit 122a. In addition, the mutation unit can also generate new individual information by using the node stored in the functional unit storage unit 10a as the mutation process. Further, it is also possible to cause the mutation portion to perform parameter mutation as a mutation process by randomly rewriting a part or all of the parameters of the node of the selected parent individual information.

Further, the gene manipulation unit 21a randomly selects the child individual information having the best evaluation value and the child individual information selected with the probability based on the ranking information of the child individual information in the evaluation result storage unit 20a from the parent population. The parent individual information in the parent population storage unit 121a can be updated by replacing the parent individual information selected in.

The image processing unit 17a inputs an image from the image storage unit 14a, inputs child individual information from the child population storage unit 122a, and sequentially executes a plurality of nodes included in the child individual information based on the child individual information. To do. Then, the image processing unit 17a creates an image as output information for each of the learning image and the verification image, and outputs this output information for each image identification information of the image and for each identification information of the child individual information. It can be stored in the storage unit 18a.

The evaluation value calculation unit 19a inputs output information from the output information storage unit 18a, and also inputs teacher information corresponding to the image identification information corresponding to the output information from the teacher information storage unit 16a.
Then, the evaluation value is calculated by comparing the output information and the teacher information, and the obtained evaluation value is stored in the evaluation result storage unit 20a for each image identification information of the image and for each identification information of the child individual information. it can.

At this time, the evaluation value calculation unit 19a calculates the evaluation value of the detection area and the evaluation value of the non-detection area of the learning image and the verification image by different calculation methods for the detection area and the non-detection area of the teacher information. In addition, the evaluation value of the individual information is calculated based on the evaluation value calculated using the learning image. Specifically, the evaluation value of the individual information can be calculated by adding up the evaluation value of the detection area and the evaluation value of the non-detection area.
At this time, the evaluation value calculation unit 19a preferably calculates the evaluation value of the detection region by point distribution analysis, and calculates the evaluation value of the non-detection region based on the brightness difference between the non-detection region and the teacher information. Is preferable.

Further, the evaluation value calculation unit 19a completes the calculation of the evaluation value for all the child individual information (or the child individual information for one generation) stored in the child population storage unit 122a in the population storage unit 12a and evaluates the evaluation value. When the result storage unit 20a stores the information, the child individual information can be ranked based on these evaluation values, and the obtained ranking information can be stored in the evaluation result storage unit 20a.
Further, the evaluation value calculation unit 19a can perform an end determination for determining whether or not to end the image processing by the image processing unit 17a.

Next, the processing procedure of the image processing apparatus of the present embodiment will be described with reference to FIG.
First, the image input unit 13a in the image processing device 1a inputs an image (step 30). Further, the teacher information input unit 15a inputs a teacher range as teacher information (step 31). As the teacher range, for example, an image showing a division between a detection area and a non-detection area can be used. Then, the individual generation unit 11a generates an initial parent population (step 32).

At this time, the individual generation unit 11a generates parent individual information by randomly selecting and arranging nodes from the functional unit storage unit 10a, and the parent individual in the population storage unit 12a for each identification information of the parent individual information. It can be stored in the group storage unit 121a. The parent population in the parent population storage unit 121a is updated by the update processing of the parent population described later, and the following processing is repeated for the set generation (step 33).

The genetic manipulation unit 21a randomly selects parent individual information for generating child individual information from the parent population in the parent population storage unit 121a. (Step 34). Then, genetic manipulation is executed on the selected parent individual information (step 35). The selection of parent individual information in step 34 may be performed as part of the genetic manipulation in step 35.

That is, the gene manipulation unit 21a executes crossing processing or mutation processing on randomly selected parent individual information to generate a plurality of child individual information, and the child population consisting of these child individual information is generated as a child. It is stored in the population storage unit 122a.

Next, the image processing unit 17a executes image processing for all the child individual information in the child population storage unit 122a (steps 36 and 37).
At this time, the image processing unit 17a can create an image or the like as output information for each child individual information and store it in the force information storage unit 18a. At this time, when the detection target exists in the learning image and the verification image, the image processing unit 17a creates an image to which the information indicating the region where the detection target exists is added and stores it in the output information storage unit 18a. Can be made to.

Next, the evaluation value calculation unit 19a calculates the evaluation value by comparing the output information and the teacher information for each child individual information (step 38), and the obtained evaluation value is stored in the evaluation result storage unit 20a as the child individual information. Remember each time.
At this time, in the calculation of the evaluation value by the evaluation value calculation unit 19a, the evaluation value of the detection region is calculated by the point distribution analysis such as the partition method. In addition, the evaluation value of the non-detection area is calculated based on the brightness difference between the non-detection area and the teacher information, and the evaluation value of the detection area and the evaluation value of the non-detection area are added up to calculate the evaluation value of the individual information. ..
Then, for all the child individual information in the child population storage unit 122a, the processes from image processing to evaluation value calculation and storage are repeatedly executed.

Next, the evaluation value calculation unit 19a ranks the child individual information based on the evaluation value stored in the evaluation result storage unit 20a (step 39).
Then, the gene manipulation unit 21a randomly selects the child individual information having the best evaluation value and the child individual information selected with the probability based on the ranking information of the child individual information in the evaluation result storage unit 20a from the parent population. The parent individual information of the parent population storage unit 121a in the population storage unit 12a is updated by replacing the parent individual information selected in (step 40).

Further, the evaluation value calculation unit 19a performs a determination process of whether or not the end condition is satisfied (step 41), and if the end condition is satisfied, the process by the image processing device 1a ends.

When the processing by the image processing device 1a is terminated, the evaluation value calculation unit 19a calculates the evaluation value for all the individual information in the final generation, ranks the individual information based on the evaluation value, and obtains the ranking information. Can be stored in the evaluation result storage unit 20a.

According to the image processing apparatus of the present embodiment as described above, genetic manipulation is executed when a parent is selected from the parent population and the child individual information is generated, and the child individual information is used as the parent individual information in the parent population. By swapping, the parent population can be updated. As a result, it is possible to search for the optimum individual information while maintaining the diversity of evolution as compared with the first embodiment.

The image processing apparatus of the above embodiment can be realized by using a computer controlled by the image processing program of the present invention. The CPU of the computer sends a command to each component of the computer based on the image processing program, and predetermined processing required for the operation of the image processing apparatus, for example, individual generation processing, image processing, evaluation value calculation processing, genetic manipulation. Have them perform processing, etc. As described above, each process and operation in the image processing apparatus of the present invention can be realized by specific means in which the program and the computer cooperate.

The program is stored in a recording medium such as ROM or RAM in advance, and the program is read by the computer from the recording medium mounted on the computer and executed. However, the program can also be read by the computer via a communication line, for example.
Further, the recording medium for storing the program can be configured by, for example, a semiconductor memory, a magnetic disk, an optical disk, or any other recording means that can be read by any computer.

As described above, according to the image processing system and the image processing program according to the embodiment of the present invention, it is possible to further optimize the image processing in the image processing technique for finding the detection target from the image.

Hereinafter, an experiment in which image processing for detecting a detection target from an image is optimized using the image processing apparatus according to the embodiment of the present invention will be described with reference to FIGS. 6 to 22.
In the examples, the image processing apparatus according to the second embodiment of the present invention was used. That is, the teacher range that separates the detected area and the non-detected area was used as the teacher information. As output information, when a detection target was detected, an image was created to which information indicating an area where the detection target exists was added. If the detection target was not detected, an image to which the information was not added was created. In addition, the evaluation value of the detection area was calculated by point distribution analysis, the evaluation value of the non-detection area was calculated by the brightness difference from the teacher information, and these were added up to obtain the evaluation value. The variance within the detection region was calculated using the equation for obtaining ^{X 2 by the interval method.} Furthermore, the genetic manipulation was carried out according to the second embodiment.

Ａ：非検出領域の面積
ｙ_Ｍ(i)：検出領域を分割した各領域において検出有りとして反応したピクセル数
Ｍ：分割数
ａ：全検出領域内で検出有りとして反応したピクセル数
ｗ：重み係数 Specifically, in the example, the evaluation value was calculated using the following formula.

A: Area of non-detection area y _M (i): Number of pixels that responded as detected in each area that divided the detection area M: Number of divisions a: Number of pixels that reacted as detected in all detection areas w: Weight coefficient

On the other hand, in the reference example, a partially modified image processing apparatus according to the second embodiment of the present invention was used. Specifically, in the reference example, the image information in which the detection target is drawn in detail for each pixel is used as the teacher information. As output information, when a detection target is detected, an image displaying the detection target is created. Further, the image information as the teacher information and the output information were compared for each pixel to calculate the mean square error, which was stored in the evaluation result storage unit as an evaluation value. In addition, the genetic manipulation was performed in the same manner as in the examples.

The test conditions of the reference example and the example are shown in FIG.
As a learning image and a verification image, 50 images each having a size of 64 × 64 pixel were prepared.
As the nodes, the input node and the output node were set to 1 node each, and the processing nodes constituting the individual information were set to 10 nodes. The nodes stored and used in the functional unit storage unit 10a are shown in FIGS. 7 to 15.

Evolutionary computation was performed assuming that the number of generations was 5000, regardless of the state of evolution. The number of parent individuals was 50, and the number of offspring was 62. The breakdown of the offspring is 2 individuals due to elite preservation, 20 individuals due to uniform crossover (change rate 20%), 20 individuals due to mutation (change rate 15%), and parameter variation (change rate 10%). ) Is 20 individuals.

The learning image, the verification image, and the teacher information were created by the steps shown in FIG.
In this embodiment, the learning image and the verification image were drawn by the program, and the teacher image was also automatically generated by the program. However, in an actual inspection scene, it is impossible to automatically generate a teacher image for various defects, so the teacher image needs to be created manually. That is, it is very complicated because it is necessary to manually create an image in which the inspection target is drawn in detail for each pixel as in the reference example, and a huge amount of work time has been required in the past.

First, the background was made (step1). For the training image and the verification image, the brightness (128 ± 25) was randomly selected and drawn for each pixel. The teacher image of the reference example and the teacher range of the example were blacked out.
Next, the non-detection target was drawn (step 2). For the learning image and the verification image, brightness (100 to 120), diameter (22 to 24 pixels), and thickness (3 to 5 pixels) were randomly selected in the center of the image, and a ring was drawn as a non-detection target. For the teacher image of the reference example and the teacher range of the example, the undetected object was not drawn.

Next, the detection target was drawn (step 3). For the learning image and the verification image, the range (15 to 25 pixels, arbitrary position), number (15 to 20), and reference brightness (128 ± 100) are randomly selected, and the diameter (for each detection target) ( 1 to 3 pixels), the change from the reference brightness (0 ± 25) was randomly selected, and a circle was drawn within the range as the detection target. For the teacher image of the reference example, a circle with the same shape as the inspection target was painted white. Regarding the teacher range of the example, the range where the detection target exists was painted white in a square shape.

Next, a process of adding noise was performed (step 4). That is, for the training image and the verification image, the brightness (0 to 50) was randomly selected with a probability of 5% for each pixel and added to the original brightness. No noise was added to the teacher image of the reference example and the teacher range of the example.
Further, a process of adding blur was performed (step 5). That is, the Gaussian filter processing (kernel size 3 × 3) was performed on the training image and the verification image. No blurring was performed on the teacher image of the reference example and the teacher range of the example.
An example of the image created in this way is shown in FIG.

As for the evaluation values, since the evaluation methods are different between the examples and the reference examples, they cannot be compared as they are. Therefore, the evaluation value was normalized using the following formula.

The following 2x2 cross table was used to compare the examples with the reference examples. In the reference example, there is no distinction between the detection area and the non-detection area, but for comparison with the example, the area in which the circle having the same shape as the inspection target is painted in white is used as the detection area, and the other areas are not. As the detection area, the number of images for which the detection target was found was totaled.

表１において、ａ、ｂ、ｃ、ｄは、以下を示す。
ａ：陽性の画像数（教師情報の検出領域において、検出対象が検出されたもの）
ｂ：偽陽性の画像数（教師情報の非検出領域において、検出対象が検出されたもの）
ｃ：偽陰性の画像数（教師情報の検出領域において、検出対象が検出されなかったもの）
ｄ：陰性の画像数（教師情報の非検出領域において、検出対象が検出されなかったもの）
感度及び特異性は、１．０が最良値となる。
なお、本実施例では、学習用画像と検証用画像のそれぞれ５０枚に対し、検出対象と非検出対象について比較しているため、ａ＋ｂ＋ｃ＋ｄ＝１００である。

In Table 1, a, b, c, d indicate the following.
a: Number of positive images (detection target detected in the detection area of teacher information)
b: Number of false positive images (detection target detected in the non-detection area of teacher information)
c: Number of false negative images (the detection target was not detected in the detection area of the teacher information)
d: Number of negative images (the detection target was not detected in the non-detection area of the teacher information)
The best value for sensitivity and specificity is 1.0.
In this embodiment, since the detection target and the non-detection target are compared for 50 images each of the learning image and the verification image, a + b + c + d = 100.

As a result of the above test, a graph showing the transition of the normalized evaluation values in the reference example and the example is shown in FIG. Further, the structure of individual information (image processing algorithm) obtained by the reference example is shown in FIG. 18, and the structure of the individual information obtained by the example is shown in FIG.
It should be noted that these evaluation values are based on the learning image. Although the evaluation values of the verification images are calculated, they are not used for comparison of the evaluation values of the reference example and the example.

First, in the reference example, the evaluation value improved sharply around the 1000th generation, and the evolution stagnated around the 1300th generation. With reference to the structure of the individual information of the reference example shown in FIG. 18, in the reference example, a complicated network for obtaining detailed output information is generated for each pixel by comparing with the teacher image. It is presumed that this is because it was necessary to consider the direction of next evolution in various directions in evolutionary computation.

On the other hand, in the examples, it can be seen that the evaluation values have been steadily improved by the 1000th generation without major stagnation. With reference to the structure of the individual information of the example shown in FIG. 19, a very simple algorithm is generated in the example. It is presumed that this is due to the fact that the direction of evolution was small. Also, the simple algorithm is because the teacher image is roughly given as a range, so unlike the reference example, it is not necessary to compare it with the teacher image in detail for each pixel, and it is roughly detected in evolutionary computation. It is thought that this is due to the fact that it was taught through the teacher image that it should be done.

As a result of the above test, FIG. 20 shows the sensitivity / specificity of the learning result and the verification result according to the reference example, and FIG. 21 shows the sensitivity / specificity of the learning result and the verification result according to the example. Further, FIG. 22 shows an example of a correct answer image and an incorrect answer image according to the reference example and the example.
Comparing the results shown in FIGS. 20 and 21, the examples showed better performance than the reference examples in both the learning result and the verification result. In addition, in the reference example, a large difference in sensitivity and specificity can be seen between the learning result and the verification result, indicating that overfitting has occurred. On the other hand, in the examples, there is no big difference between the learning result and the verification result, and it can be seen that overfitting is suppressed.

Further, referring to FIG. 22, it can be seen that the reference example attempts to detect the detection target in detail, and the embodiment detects only a characteristic portion of the detection target. The reason is that, as described above, in the examples, the teacher information is given as if the detection target should be roughly detected, and this is reflected in the evolutionary computation, and the performance difference from the reference example. Is presumed to have given birth to. In both the reference example and the example, the central ring that was not detected was not detected, and the background noise was erroneously detected.

As described above, it was found from this test that the image processing device of the example in which the teacher information can be easily created has better detection performance than the image processing device of the reference example. It was also clarified that the image processing apparatus of the embodiment can significantly reduce the time and effort required to create teacher information.

It goes without saying that the present invention is not limited to the above embodiments, and various modifications can be made within the scope of the present invention.
For example, each configuration in the image processing device can be distributed to a plurality of information processing devices, or can be used as a node including a node other than the one used in the embodiment.
Further, the above image processing device is used for image inspection, the image input unit inputs an image for inspection, and the image processing unit processes the image for inspection based on the individual information having the highest evaluation value. It is possible to make appropriate changes such as a configuration including a processing result storage unit for storing the processing result information obtained in the above process.

INDUSTRIAL APPLICABILITY The present invention is a case where inspection or the like is performed based on image data of equipment or a product, and can be suitably used for obtaining an information processing apparatus for image inspection with optimized image processing. ..

1,1a Image processing device 10,10a Functional unit storage unit 11,11a Individual generation unit 12, 12a Individual group storage unit 121a Parent population storage unit 122a Child population storage unit 13, 13a Image input unit 14, 14a Image storage unit 15,15a Teacher information input unit 16,16a Teacher information storage unit 17,17a Image judgment unit 18,18a Output information storage unit 19, 19a Evaluation value calculation unit 20, 20a Evaluation result storage unit 21,21a Gene manipulation unit

Claims (11)

It is an image processing system that optimizes image processing that detects the detection target from an image by using genetic manipulation, which is an optimization method that mathematically simulates the evolution of living things.
A population storage unit that stores multiple individual information that defines the execution order of multiple functional units in a network that can include a cycle, and a population storage unit.
An image storage unit that stores a plurality of learning images and a plurality of verification images,
A teacher information storage unit that stores teacher information that separates a detection area and a non-detection area for evaluating output information, which is associated with the learning image and the verification image, respectively.
An image processing unit that processes the learning image and the verification image based on the individual information and generates the output information.
An evaluation value calculation unit that calculates the evaluation value of the individual information by comparing the output information with the teacher information and ranks the individual information based on the evaluation value calculated using the learning image.
A gene manipulation unit that updates the population storage unit by selecting or changing the individual information based on the ranking information of the individual information is provided.
The evaluation value calculation unit calculates the evaluation value of the detection region and the evaluation value of the non-detection region of the learning image and the verification image by different calculation methods for the detection region and the non-detection region of the teacher information. An image processing system characterized by calculating and calculating the evaluation value of the individual information based on these evaluation values. The image processing system according to claim 1, wherein the evaluation value of the detection region is calculated by point distribution analysis. The image processing system according to claim 2, wherein the point distribution analysis is at least one of a K-function method, a nearest neighborhood distance, and a partition method. The image processing system according to any one of claims 1 to 3, wherein the evaluation value of the non-detection region is calculated based on the brightness difference from the teacher information. The processing according to any one of claims 1 to 4, wherein the processing by the image processing unit includes at least one of noise removal, luminance correction, binarization, edge detection, frequency filter, and four-rule arithmetic processing. Image processing system. The processing by the image processing unit is characterized in that it includes a process of calculating a feature amount in a certain range in the image and a process of determining whether or not the certain range is a detection target based on the feature amount. The image processing system according to any one of claims 1 to 5. The genetic manipulation unit
The elite storage unit that leaves individual information with high evaluation values to the next generation,
Based on the evaluation value, a selection unit that leaves individual information to the next generation with a certain probability,
A crossing processing unit that exchanges a part of two individual information with each other,
The image processing system according to any one of claims 1 to 6, further comprising a mutant portion that randomly rewrites a part or all of one selected individual information. The image input unit inputs an image for inspection, and the processing result storage unit stores the processing result information obtained by processing the image for inspection by the image processing unit based on the individual information having the highest evaluation value. The image processing system according to any one of claims 1 to 7, wherein the image processing system is provided. An image processing program that optimizes image processing to detect a detection target from an image using genetic manipulation, which is an optimization method that mathematically simulates the evolution of living organisms.
Computer,
A population storage unit that stores multiple individual information that defines the execution order of multiple functional units in a network that can include a cycle.
Image storage unit that stores learning images and verification images,
A teacher information storage unit that stores teacher information that separates a detection area and a non-detection area for evaluating output information, which is associated with the learning image and the verification image, respectively.
An image processing unit that processes the image based on the individual information and generates the output information.
An evaluation value calculation unit that calculates an evaluation value by comparing the output information with the teacher information and ranks the individual information based on the evaluation value calculated using the learning image, and an evaluation value calculation unit.
The individual information is selected or changed based on the ranking information of the individual information to function as a gene manipulation unit for updating the population storage unit.
The evaluation value calculation unit is provided with an evaluation value of the detection area and an evaluation value of the non-detection area of the learning image and the verification image by different calculation methods for the detection area and the non-detection area of the teacher information. An image processing program for calculating and calculating the evaluation value of the individual information based on these evaluation values. The image processing program according to claim 9, wherein the evaluation value of the detection region is calculated by point distribution analysis. The image processing program according to claim 9 or 10, wherein the evaluation value of the non-detection region is calculated based on the brightness difference from the teacher information.

Images