KR20230063859A - Parameter adjustment support device and parameter adjustment support method - Google Patents

Abstract

The present invention provides a technique for simplifying parameter adjustment work by a user for an image sensor having a plurality of test processes. A parameter adjustment support device comprises: an adjustment target setting part setting test processes of N items to be adjusted among a plurality of test processes; an image acquisition part acquiring an OK image and an NG image; a graph generation part generating a graph showing an OK image measurement value which is the value of a measurement parameter calculated in accordance with a current condition of a setting parameter from the OK image for each of the test processes of the N items, an NG image measurement value which is the value of a measurement parameter calculated in accordance with the current condition of the setting parameter from the NG image of the corresponding test process, and a current threshold; and an output part outputting the graph to a display device.

Description

Parameter adjustment support device and parameter adjustment support method {PARAMETER ADJUSTMENT SUPPORT DEVICE AND PARAMETER ADJUSTMENT SUPPORT METHOD}

The present invention relates to a technique for supporting adjustment of parameters used in an inspection process of an image sensor.

BACKGROUND OF THE INVENTION [0002] A system called an image sensor is frequently used in a production line of a factory or the like to automate or save labor for product inspection (see Patent Document 1). In the image sensor, a series of inspection flows are configured by combining inspection processing of a plurality of items. In order to obtain correct inspection results with the image sensor, it is necessary to appropriately set the parameters used in each inspection process, but conventionally, there is no other way but to individually set the parameters of each inspection process, and adjustment (alignment) of the parameters was not an easy task.

Japanese Unexamined Patent Publication No. 2008-015706

The present invention has been made in view of the above situation, and its object is to provide a technique for simplifying the adjustment work of parameters by a user regarding an image sensor having a plurality of inspection processes.

The present disclosure relates to an image sensor capable of performing a plurality of inspection processes on an image, and a parameter adjustment support device for supporting adjustment of parameters used in each inspection process, wherein the inspection process determines a predetermined measurement parameter from an image. A value is calculated, and a value of the measurement parameter is compared with a threshold value to output an OK or NG determination result, wherein the parameter includes a setting parameter defining conditions for calculating the value of the measurement parameter and the threshold value It is necessary to obtain an OK determination in all of the adjustment target setting unit for setting the inspection processing of N items (N is an integer of 2 or more) to be adjusted among the plurality of inspection processes, and the inspection processing of the N items. An image acquisition unit that acquires an OK image to be performed and an NG image for which an NG judgment is to be obtained in at least one of the N-item inspection processes, and for each of the N-item inspection processes, from the OK image an OK image measurement value that is the value of the measurement parameter calculated according to the current condition of the setting parameter, and an NG image measurement value that is the value of the measurement parameter calculated from the NG image of the inspection process according to the current condition of the setting parameter; A parameter adjustment support device characterized in that it includes a graph generating unit for generating a graph in which the current threshold value is drawn, and an output unit for outputting the graph to a display device.

The OK image measurement value, the NG image measurement value, and the threshold value may be drawn on the graph in a manner in which at least one of color or graphic is different.

The graph generating unit determines whether adjustment of the parameter of the inspection process is necessary based on a relationship between the OK image measurement value, the NG image measurement value, and the threshold value in the inspection process for the adjustment target, and The judgment result may be drawn on the graph.

The graph generating unit may further determine which adjustment of the threshold value and the setting parameter is required, and may plot the result of the determination on the graph.

The graph generating unit determines at least three states among the parameter adjustment unnecessary, the threshold adjustment necessary, and the setting parameter adjustment necessary, even if the three states are drawn in the graph in different colors. do.

The output unit may output an adjustment window for performing an adjustment operation of the parameters to the display device.

In the adjustment window, a scatter diagram in which the OK image measurement values of each OK image, the NG image measurement values of each NG image, and the threshold values are plotted may be displayed.

The adjustment window sets the OK image measurement value and the NG image measurement value drawn on the scatter plot to the OK image measurement value of an OK image outside the OK range defined by the threshold value and the NG image of an NG image included in the OK range. It may be possible to switch only with the image measurement value.

The present disclosure is a parameter adjustment support method for supporting adjustment of parameters used in each inspection process with respect to an image sensor capable of executing a plurality of inspection processes on an image, wherein the inspection process determines a predetermined measurement parameter from an image. A value is calculated, and a value of the measurement parameter is compared with a threshold value to output an OK or NG determination result, wherein the parameter includes a setting parameter defining conditions for calculating the value of the measurement parameter and the threshold value The step of setting the inspection process of N items (N is an integer of 2 or more) to be adjusted among the plurality of inspection processes, and the OK image for which OK determination is to be obtained in all of the inspection processes of the N items. and a step of acquiring an NG image for which an NG judgment is to be obtained in at least any one of the N items of inspection processing, and for each of the N items of inspection processing, a current value of the setting parameter from the OK image. An OK image measurement value that is the value of the measurement parameter calculated according to conditions, an NG image measurement value that is the value of the measurement parameter calculated according to the current condition of the setting parameter from the NG image of the inspection process, and the current threshold A method for supporting parameter adjustment comprising a step of generating a drawn graph and a step of outputting the graph to a display device.

The present disclosure includes a program for causing a processor to execute each step of the method for supporting parameter adjustment.

The present invention may be regarded as a parameter adjustment support device having at least a part of the above means, or as an image sensor having the same device. In addition, the present invention can also be understood as a parameter adjustment support method or image sensor control method including at least a part of the above processing, or a program for realizing such a method, or a recording medium on which the program is recorded non-temporarily. Further, each of the above means and processing can be combined with each other as much as possible to constitute the present invention.

According to the present invention, it is possible to simplify the adjustment operation of the parameters by the user regarding the image sensor having a plurality of inspection processes.

1 is a diagram schematically showing the appearance of an image sensor.
Fig. 2 is a functional block diagram showing a configuration example of an inspection function and a parameter adjustment support function.
Fig. 3 is a diagram for explaining the process flow of inspection and parameter setting.
4 is a flowchart showing the flow of a parameter adjustment operation.
5 is an example of the main screen of the parameter adjustment support tool.
6 is an example of the label setting tab of the image classification window.
7 is an example of a labeling tab of an image classification window.
8 is an example of a radar chart display.
Fig. 9 is an example of adjusting the threshold by an adjustment window.
Fig. 10 is an example of adjusting setting parameters using an adjustment window.
Fig. 11 is an example of adjusting the setting parameters of the area center of gravity.
Fig. 12 is an example of adjusting setting parameters using an adjustment window.
13 is an example of a radar chart after completion of parameter adjustment.

Referring to Fig. 1, an image sensor according to an embodiment of the present invention will be described. 1 is a diagram schematically showing the appearance of an image sensor.

The image sensor 1 is a device installed in a manufacturing line PL of a factory, for example, and used for various processes using images. The image sensor 1 is also called an image processing system, a vision sensor, a vision system, or the like.

The image sensor 1 includes an imaging device 10, an image processing device 11, a display device 12, and an input device 13 as hardware resources. The imaging device 10 , display device 12 , and input device 13 are connected to the input/output I/F of the image processing device 11 . 1 shows the image sensor 1 in which the imaging device 10 and the image processing device 11 are separate structures, however, a processing-integrated image sensor in which the imaging device and the image processing device are integrated may be used.

The imaging device 10 is a device including a lighting unit, a lens unit, an imaging unit, and the like, and is also called an industrial camera. The lighting unit is a device that illuminates the subject (object to be inspected ○), and is composed of, for example, an LED light source or the like. The lens unit is an optical device that forms an optical image of a subject on the image pickup unit, and an optical system having functions such as focus adjustment, diaphragm, and zoom is used. An imaging unit is a device that generates and outputs image data by photoelectric conversion, and is constituted by, for example, an imaging element such as a CCD or CMOS sensor.

The image processing device 11 has, as its main functions, introduction of image data from the imaging device 10, image processing on the image data, inspection processing based on the result of image processing, and communication with an external device through an input/output I/F. It has data transmission/reception, generation of data to be output to an external device, processing of data received from an external device, control of the imaging device 10 and input/output I/F, and the like. The image processing device 11 includes, for example, a processor (CPU, GPU, etc.), memory (RAM, ROM, etc.), storage (non-volatile storage devices such as hard disks and solid state drives), input/output I/F, etc. It can be configured as a computer for image processing having a. The functions and processes of the image processing device 11 described later can be realized by loading a program stored in storage into a memory and executing it by a processor. However, some or all of the functions of the image processing device 11 may be implemented by an Application Specific Integrated Circuit (ASIC) or a Field Programmable Gate Array (FPGA), or may be executed by another computer, cloud server, or the like.

FIG. 2 is a functional block diagram showing a configuration example of an inspection function and a parameter adjustment support function in the image processing device 11. As shown in FIG.

The image processing device 11 includes an image input unit 20, an image storage unit 21, an inspection execution unit 22, a parameter storage unit 23, an inspection process setting unit 24, and a parameter adjustment support unit 25. do. The image input unit 20 introduces image data from the imaging device 10 . The introduced image data is stored in the image storage unit 21. The inspection execution unit 22 performs inspection of a plurality of items on the image data, and outputs an OK/NG determination result for each item. The judgment result of the inspection may be linked to image data and stored. In the parameter storage unit 23, setting values of parameters used in each inspection process are registered. The inspection execution unit 22 refers to parameter setting values from the parameter storage unit 23 when executing inspection processing. The inspection process setting unit 24 provides a user interface (UI) for setting parameters used in the processing flow of inspections performed by the inspection execution unit 22 and individual inspection processes. The parameter adjustment support unit 25 provides a tool for assisting a user to adjust (fit) a parameter to an appropriate value. The parameter adjustment support unit 25 has an image acquisition unit 250, an adjustment target setting unit 251, a graph generation unit 252, and a UI output unit 253. Details of these functions will be described later.

(Inspection processing settings)

Referring to Fig. 3, an example of a procedure for setting the processing flow and parameters of the inspection using the UI of the inspection process setting unit 24 will be described.

In the image processing device 11, a plurality of types of modules (hereinafter referred to as "processing units") that execute specific processing are prepared. The processing units are roughly divided into those related to inspection processing and those related to processing other than inspection processing. As the processing unit for inspection processing, for example, "shape search" for inspecting whether or not a region of a predetermined shape exists in an image, "center of gravity of area" for examining the area or position of a region of a predetermined color, and the position of an edge of a subject Various processing units are prepared, such as "edge position" that inspects . In addition, as processing units other than inspection processing, for example, a "camera image input" that introduces a captured image from the imaging device 10, a "color extraction filter" that extracts a predetermined color range from an image, etc. are prepared. . As shown in FIG. 3 , the user sets the inspection to be performed by appropriately combining processing units using the UI of the inspection process setting unit 24 . A series of processing created by combining a plurality of processing units is referred to as a “processing flow”.

The processing unit of the inspection processing calculates the value of a predetermined measurement parameter based on the feature extracted from the image, compares the value of the measurement parameter with a threshold value, and outputs a determination result of "OK" or "NG". has the design of For example, in the case of the above-mentioned "shape search", as measurement parameters, the degree of similarity between the registered model and the model detected from the image (correlation value), the position where the model was detected (XY coordinates), the inclination of the detected model ( angle relative to the registration model), etc. are calculated. Then, for each of the similarity, position, and gradient, if threshold values (threshold upper limit and lower threshold value defining the OK range) are set, and all values of the similarity, position, and gradient are included in the OK range, “OK”, 1 If even a dog is out of the OK range, a judgment result of "NG" is output. In addition, in the case of "Area Center of Gravity", the area of the region, the position of the center of gravity of the region (XY coordinates), etc. are calculated as measurement parameters. Then, a threshold value is set for each of the area and the center of gravity position, and if both the area and the center of gravity position are included in the OK range, a judgment result of “OK” is output, and if either is out of the OK range, “NG” is output. . The "shape search" and "area center of gravity" exemplified here are specifications for making OK/NG decisions for multiple types of measurement parameters, that is, specifications for performing inspection processing on multiple items in one processing unit. There is also a processing unit that has only one type of measurement parameter (one item of inspection process).

Indicators used as measurement parameters in each processing unit are predefined and cannot be changed by the user. However, the user can set about various conditions at the time of calculating the value of a measurement parameter (hereinafter referred to as "setting parameter") and a threshold value used for determination. For example, as setting parameters in "shape search", a model to be searched, the position of the model, and a feature point on the model to be paid attention to in calculating the degree of similarity, and the like can be cited. In addition, as shown in Fig. 3, as the setting parameter in the "Area Center of Gravity", the color condition (each range of hue, saturation, and brightness) to be extracted as a region for calculating the area or center of gravity, etc. can be cited. there is. Depending on the setting of these setting parameters, the value of the measurement parameter calculated from the image changes greatly. In addition, whether the threshold value is appropriately set affects the OK/NG determination accuracy and the reliability of the determination result.

Therefore, after the user creates a processing flow by combining processing units, it is necessary to appropriately set the setting parameters and threshold values of each processing unit so that desired inspection performance (determination result) is obtained. However, a high degree of skill and know-how is required for the operation of adjusting (setting) the setting parameters and thresholds to appropriate values. In addition, when the number of processing units included in the processing flow is large, or when each processing unit has a large number of setting parameters or thresholds, adjusting the setting parameters and thresholds in a well-balanced manner so as to obtain appropriate judgment results for all of them is difficult. , not easy.

Therefore, in the image sensor 1 of the present embodiment, a parameter adjustment support tool for supporting adjustment work of setting parameters and threshold values is provided by the parameter adjustment support unit 25 (this parameter adjustment support tool is not included in the present invention). It is an example of a parameter adjustment support device for

(Parameter adjustment support tool)

Referring to Figs. 4 to 12, the parameter adjustment support tool of the present embodiment will be described in detail. Fig. 4 is a flowchart showing the flow of a parameter adjustment operation using the parameter adjustment support tool. 5 is an example of the main screen of the parameter adjustment support tool.

When the parameter adjustment support tool (hereinafter also simply referred to as "tool") is activated, the main screen in Fig. 5 is displayed on the UI output unit 253 (step S40). On the main screen, a list 50 of a plurality of processing units constituting the processing flow is displayed.

The user first designates the location where the image (sample image) used for the adjustment work is stored in the column 51 of the "re-measurement target image" of the main screen (step S41). When a "body logging image" is designated, a group of images stored in the image storage unit 21 (for example, a RAM disk or internal storage of the image sensor 1, etc.) is used, and a "specified folder image" is designated. In this case, a group of images stored in an external storage can be used. When the storage location is specified, the sample image is imported by the image acquiring unit 250 . Although it is possible to perform the adjustment operation with only one sample image, a plurality of sample images obtained by photographing different objects to be inspected may be used for the adjustment operation. In addition, the plurality of sample images need only include both images of good products and images of defective products, and among the images of defective products, a plurality of types of defects (that is, different inspection processing items resulting in NG judgment) are included. It should be. By simultaneously and progressively performing parameter adjustment for a plurality of inspection processing items using a plurality of images with variations, it is possible to efficiently perform an appropriate parameter matching operation.

Next, when the user presses the "Image Classification" button 52 on the main screen, the adjustment target setting unit 251 displays the image classification window shown in Fig. 6 . The user designates inspection processing as an adjustment target in the "label setting" tab 60 of the image classification window, and performs an operation of setting labels for them (step S42). Specifically, the number of items of the inspection process to be adjusted is entered in the "number of settings" column 61. The number of items N can be set within the range of 3 to 20, for example. When the number of items N is input, N labels are generated, so the user sets "label name", "target unit", and "measurement parameter" for each label. "Label name" can be set arbitrarily. In "Target Unit", the name of the processing unit that performs the inspection process to be adjusted is specified, and in "Measurement Parameter", the name of the measurement parameter used in the inspection process to be adjusted is specified. Fig. 6 shows an example in which five labels of "Circularity", "Printing Area", "Presence of Printing", "Printing Quality", and "Printing Height" are set.

When the label setting is finished, the user performs an operation of assigning a label to the image in the "Labeling" tab 62 (step S43). When the "labeling" tab 62 of the image classification window is selected, the UI shown in Fig. 7 is switched. The UI of the "Labeling" tab 62 includes a label display 71, a preview button 72, a file list 73, and a labeling column 74. In the label display 71, a list of "judgment label" and "defect type label" is displayed. The judgment label is a label indicating the judgment result of the image, and three types of labels of "OK", "NG", and "Classification Suspended" are prepared in advance. The defective type label is a label defined by the user in the "label setting" tab 60. In the file list 73, a list of file names of the images designated in step S41 is displayed. When a plurality of images are designated in step S41, the file names of the plurality of images are displayed in the file list 73. When the preview button 72 is pressed, a preview of an image selected in the file list 73 can be displayed. The user first assigns a judgment label while confirming the preview image. An “OK” label is applied to images to be judged OK in all inspection processes to be adjusted, an “NG” label is applied to images to be judged NG in any of the inspection processes, and “Sorting Pending” is applied to images for which neither OK nor NG judgment can be made. give a label In the label assignment column 74, cells corresponding to respective labels are prepared, and assignment and release of labels can be switched each time the cell is pressed. For the image to which the "NG" label has been assigned, a defective type label is subsequently also assigned. That is, information that should be determined as NG in a certain inspection process is associated with an image. It is also possible to assign a plurality of defect type labels to one image. In the label display 71 and the label assignment column 74, each label may be displayed in a different color. This makes it easier to visually recognize the label attached to each image, and easier to perform the labeling operation, as well as preventing operation errors such as assigning an incorrect label. By performing the above operations, a plurality of OK images and a plurality of NG images can be prepared.

When labeling (i.e., classification of images) is completed and the OK button is pressed, the screen returns to the main screen. In addition, when it is desired to save the content set in the "label setting" tab 60 or the "labeling" tab 62, it is possible to save it in a file in the "data preservation" tab 63.

Next, when the user presses the "execute batch remeasurement" button 53 on the main screen, the inspection execution unit 22 performs a plurality of inspection processes for each of the classified images (labeled images) according to the processing flow. is performed, and measurement parameters are calculated and threshold values are determined in each inspection process (step S44). At this time, the current condition value and the current threshold value of the setting parameters stored in the parameter storage unit 23 are used.

After the re-measurement process in step S44 is completed, when switching to the "data details" tab 54 of the main screen, a graph showing the result of the re-measurement process is generated by the graph generating unit 252 (step S45), and the UI A graph is displayed on the main screen by the output unit 253 (step S46). In the graph, for example, for each of the inspection processes to be adjusted (that is, the inspection process in which the defective type label is set), "Value of the measurement parameter calculated from the OK image according to the current condition of the setting parameter", "NG" The value of the measurement parameter calculated according to the current condition of the setting parameter" and the "current threshold value" should just be drawn from the image. The specific form of the graph is not limited, but any form in which the results of the inspection process for a plurality of items can be confirmed at a glance is sufficient. For example, radar charts, bar graphs, and line graphs can be used.

8 shows an example of a radar chart display. Each axis of the radar chart corresponds to an inspection process in which a defective type label is set. For example, as shown in FIG. 6, in the case of setting five defective labels of "Circularity", "Printing Area", "Presence of Printing", "Printing Quality", and "Printing Height", as shown in FIG. The radar chart becomes a pentagon.

On each axis of the radar chart, there are eight points: "Max OK", "Mean OK", "Minimum OK", "Max NG", "Mean NG", "Minimum NG", "Upper Threshold", and "Lower Threshold". portrayed The maximum OK value, the average OK value, and the minimum OK value indicate the maximum value, average value, and minimum value of measurement parameter values respectively calculated from a plurality of OK images. Further, the NG maximum value, the NG average value, and the NG minimum value represent the maximum value, average value, and minimum value of measurement parameters calculated from a plurality of NG images, respectively. In this way, eight points of "OK maximum value", "OK average value", "OK minimum value", "NG maximum value", "NG average value", "NG minimum value", "Threshold upper limit value", and "Threshold lower limit value" are set for each inspection process (axis By presenting to the user a graph drawn in association with each), it is possible to facilitate understanding of the trend of measurement parameters in each inspection process and the relationship with threshold values. Here, the measurement parameter values calculated from the OK image (OK maximum value, OK average value, OK minimum value) and the measurement parameter values calculated from the NG image (NG maximum value, NG average value, NG minimum value) and threshold values (threshold upper limit, threshold lower limit) It is only necessary that at least one of the color or figure is drawn in a different manner. Further, the maximum value, the average value, and the minimum value may be drawn in a manner in which at least one of the colors or graphics is different. By adopting such a display mode, OK, NG, and threshold values can be visually distinguished easily.

The scale of each axis of the radar chart may be standardized using the minimum and maximum values among the above eight values. That is, the scale may be adjusted so that the minimum and maximum values of the eight values come to or near both ends of each axis of the radar chart (the center and the outermost end of the chart). As a result, since the plot positions of the eight points are enlarged as much as possible, the visibility (easiness of viewing) is improved.

On each axis of the radar chart, the "label name" of the defective type label and the "status of setting adjustment" are displayed. There are three patterns of the setting adjustment state: "setting completed", "threshold adjustment required", and "setting adjustment required". The graph generation unit 252 calculates the measurement parameter values calculated from the OK image (hereinafter simply referred to as "OK image measurement values") and the measurement parameter values calculated from the NG images (hereinafter simply referred to as "NG image measurement values") The state of setting adjustment is judged based on the relationship between ? and the threshold value. In the example of FIG. 8 , the state of setting adjustment is indicated by color icons (for example, green: setting completed, yellow: threshold adjustment required, red: setting adjustment required).

Here, "set completed" is a state in which the OK image measurement value and the NG image measurement value are separated, and the threshold value is set between the OK image measurement value and the NG image measurement value. If this state is obtained, it can be expected that an appropriate OK/NG determination is performed in the inspection process, so it may be determined that the setting of the setting parameters and threshold values related to the measurement parameter is completed (no adjustment required). In the example of Fig. 8, three of "Circularity", "Printing Quality", and "Presence/absence of Printing" are in the "Setting Completed" state.

"Threshold adjustment is required" is a state in which the OK image measurement value and the NG image measurement value are separated, but the threshold value is not set between the OK image measurement value and the NG image measurement value. In the example of Fig. 8, the "printing height" is in the "threshold adjustment required" state. In this case, the threshold value is adjusted so that the threshold value is between the measured value of the OK image and the measured value of the NG image, so that the state of "setup is completed".

"Need to adjust settings" is a state in which OK image measurement values and NG image measurement values are not separated. In this case, since adjustment of the threshold alone cannot solve the problem, adjustment of setting parameters is required. In the example of Fig. 8, the "printing area" is in the "setting adjustment required" state.

By displaying the setting adjustment state together with a graph, the user can distinguish between those with appropriate parameter settings (no adjustment required) and those requiring parameter adjustment, and further parameter adjustments, for the plurality of inspection processes selected as adjustment targets. Regarding this necessity, it is possible to immediately and reliably determine whether adjustment of the threshold is sufficient or whether the setting parameters must be adjusted. Further, in the graph, not only the determination result (color icon) of "status of setting adjustment" is displayed, but also the numerical relationship between the measured value of the OK image, the measured value of the NG image, and the threshold value. Therefore, it is possible to easily check the basis for determination (evidence) of the "status of setting adjustment", and the user's persuasiveness can be improved. In addition, since the user can grasp how or to what extent the threshold value or setting parameter should be adjusted from the numerical relationship between the measured value of the OK image, the measured value of the NG image, and the threshold value, the subsequent parameter adjustment work can be efficiently performed. it becomes possible Using not only one image, but also multiple OK images and multiple NG images, “OK maximum value”, “OK average value”, “OK minimum value”, “NG maximum value”, “NG average value”, and “NG minimum value” are plotted on a graph. Since it can be displayed, the user can grasp the adjustment status of parameters for a plurality of images as a whole, and it becomes possible to efficiently perform parameter adjustment work.

If the inspection process (label) determined as "needs threshold adjustment" or "needs setting adjustment" remains (NO in step S47), the user proceeds with the task of adjusting the threshold of the test processing or the adjustment of the setting parameters. is performed (step S48).

Referring to Fig. 9, an operation in the case of "threshold adjustment required" will be described. When a label name displayed on the radar chart in Fig. 8 (for example, "printing height") is selected (presses), the UI output unit 253 displays an adjustment window for adjusting parameters related to the label. (FIG. 9).

A scatter plot 90 and a histogram 91 show details of the distribution of OK image measurement values and NG image measurement values for the selected label. The horizontal axis of the scatter plot 90 corresponds to individual images (image No.), and the vertical axis corresponds to measured values (values of measurement parameters calculated from images). The horizontal axis of the histogram 91 is the frequency (number of images), and the vertical axis is the same as the vertical axis of the scatter plot 90. Here, in the scatter plot 90, the arrangement of the images on the horizontal axis can be determined so that the group of OK images and the group of NG images are divided (in the example of FIG. 9, the OK images are arranged first, and the NG images are arranged in the second half). are arranged, and a line representing the boundary between the OK image group and the NG image group is drawn). In addition, the scale of the vertical axis may be adjusted so that the minimum and maximum values of the measured values are appropriately separated. In addition, in the scatter plot 90 and the histogram 91, it is good if the drawing color is different between the OK image and the NG image. By such a study, it becomes easy to grasp the distribution and tendency of OK image measurement values and the distribution and tendency of NG image measurement values.

In the lower text boxes 92B and 92U of the scatter plot 90, the current set values of the threshold (lower limit) and threshold (upper limit) are displayed. Further, lines 93B and 93U indicating the threshold value (lower limit) and threshold value (upper limit) are drawn on the scatter plot 90 and histogram 91.

As can be seen from Fig. 9, although the OK image measurement values and NG image measurement values are well separated, both the OK image measurement value and the NG image measurement value have a threshold value (lower limit) and a threshold value (upper limit). It falls within the OK range defined by . In such a case, the user can adjust the threshold value by changing the values of the text boxes 92B and 92U or dragging and moving the lines 93B and 93U. In the example of Fig. 9, by correcting the threshold value (upper limit) to fall between the OK image measurement value and the NG image measurement value, an appropriate judgment result is obtained. It is also possible to automatically adjust the threshold. When the "auto adjustment" button 94 is pressed, an appropriate threshold value is automatically calculated based on the measured value of the OK image and the measured value of the NG image. The algorithm for determining the threshold is arbitrary. For example, the threshold value may be determined so as to be exactly halfway between the representative values (average value, median value, minimum value, maximum value, etc.) of OK image measurement values and representative values (average value, median value, minimum value, maximum value, etc.) of NG image measurement values. Alternatively, a threshold determination algorithm such as discriminant analysis may be used.

In addition, although FIG. 9 shows an example of dozens of images, it is also assumed that parameter adjustment is performed using images on the order of hundreds or thousands in actual operation. If the number of images becomes large, the visibility of the scatter plot 90 or the histogram 91 may be reduced, and there is a risk that an image in which an erroneous judgment has occurred may be overlooked or an appropriate arrangement of threshold values may be mistakenly viewed. In such a case, the "adjust settings in misjudgment image" button 95 is useful. When this button 95 is pressed, only images in which erroneous judgment has occurred (i.e., OK images whose measured values deviate from the OK range and NG images whose measured values fall within the OK range) are extracted from the population of images, and the extracted Only information on erroneous judgment images is displayed on the scatter plot 90 and histogram 91. By switching to such a display, it is possible to check the state of the measured values for each image while viewing the measured values for a plurality of images, and since the problematic location becomes clear, efficient and appropriate parameter adjustment becomes possible.

The user can select one image by selecting a plot point on the scatter plot 90 or by inputting an image No. in the remeasurement image column 96 . When an image is selected, the measured value of the selected image is displayed on a scatter plot 90. In addition, the inspection process (remeasurement) of each processing unit is executed for the selected image, and the OK/NG judgment result is displayed in the processing unit list 97, and the judgment result and measurement of the processing unit subject to adjustment are displayed. The value is displayed in the detailed display column 98. By using these functions, it is possible to easily check measurement values or judgment results for individual images.

Referring to Fig. 10, the operation in the case of "requires setting adjustment" will be described. Fig. 10 is an example of an adjustment window displayed when the "printing area" label is selected in the radar chart of Fig. 8 .

Looking at the scatter plot 90 and the histogram 91, it can be seen that the measured values of the OK image and the measured values of the NG image are not separated and cannot be solved only by adjusting the threshold. In such a case, it is sufficient to pay attention to a group of OK images and NG images for which similar measured values have been obtained, or to focus on images for which measured values significantly different from other OK images have been obtained, and adjust setting parameters. In the example of Fig. 10, the OK image of No. 12 has the same measured value as the NG image of No. 15, and the measured value is clearly smaller than that of the other OK images. Therefore, it can be determined that performing parameter adjustment corresponding to the OK image of No. 12 is a shortcut.

Therefore, the user selects the OK image of No. 12 on the scatter plot 90 or inputs "12" into the remeasured image column 96 to select the image of No. 12. Under the current conditions, "NG" is displayed as the determination result of the corresponding processing unit "area center of gravity", and a measured value of "area: 0.0000" is displayed. Looking at the scatter plot 90, it can be inferred that there is a problem with the setting parameter for calculating the area, since the measurement value of the area in the other OK images is approximately 10000.

In order to modify setting parameters, it is sufficient to select (press) a processing unit in the processing unit list 97. For example, if “area center of gravity” is selected, as shown in FIG. 11, a setting window for setting setting parameters for area center of gravity is displayed. Since nothing is displayed in the preview on the right side of the setting window, it can be seen that extraction of the area has failed in the current color designation settings (hue, saturation, brightness). Therefore, while checking the preview, the user adjusts the range of hue, saturation, and brightness so that the area to be inspected is just extracted. The lower part of FIG. 11 shows that a region can be extracted by changing the upper limit value of brightness to 255. When the setting parameter adjustment is completed, press the "OK button" to close the setting window.

As shown in Fig. 12, when returning to the adjustment window again and pressing the "Execute batch remeasurement" button 99, the measurement parameters are recalculated using the set parameters after adjustment. In the example of Fig. 12, it can be seen that the measured value of image No. 12 is 9631, and a value equal to that of the other OK images is obtained. In this state, it is possible to set a threshold for separating OK image measurement values and NG image measurement values. Therefore, the adjustment of the threshold value is continuously performed, and the adjustment work is completed.

In the procedure described above, after adjusting the parameters for both the label of "Need to adjust threshold" and the label of "Need to adjust settings", press the "Execute batch remeasurement" button 53 on the main screen, and Update the chart. As shown in Fig. 13, if the status of all the labels is "Setting complete", the parameter adjustment operation is completed (Yes in step S47 of Fig. 4).

The above embodiments are merely illustrative of configuration examples of the present invention. The present invention is not limited to the above specific forms, and various modifications are possible within the scope of the technical idea. For example, although a radar chart was exemplified in the above embodiment, a bar graph, a broken line graph, or the like may be used. Further, instead of plotting the representative values (maximum value, average value, minimum value) of OK image measurement values and NG image measurement values on a radar chart or the like, all individual measurement values may be plotted. In addition, the UI and parameters exemplified in the above embodiment are merely examples, and may be modified appropriately. Further, in the above embodiment, the user selects the inspection process to be adjusted, but the parameter adjustment support tool automatically selects (sets) the inspection process to be adjusted or the inspection to be adjusted. The processing may be requested by the user. For example, among a plurality of inspection processes included in the processing flow, N inspection processes may be selected as an inspection target (or a recommendation target) in order from the largest number of NG determination images. Alternatively, the inspection process to be adjusted may be determined in advance.

Claims (10)

A parameter adjustment support device for supporting adjustment of parameters used in each inspection process for an image sensor capable of executing a plurality of inspection processes on an image,
The inspection process calculates a value of a predetermined measurement parameter from an image, compares the value of the measurement parameter with a threshold value, and outputs an OK or NG determination result;
The parameter includes a setting parameter that defines conditions for calculating the value of the measurement parameter and the threshold value,
an adjustment target setting unit configured to set inspection processes of N items (N is an integer of 2 or greater) to be adjusted among the plurality of inspection processes;
an image acquiring unit that acquires an OK image for which an OK determination is to be obtained in all of the N items of inspection processing and an NG image for which an NG judgment is to be obtained in at least one of the N items of inspection processing;
For each of the N inspection processes, an OK image measurement value, which is a value of the measurement parameter calculated from an OK image according to the current condition of the setting parameter, and a current condition of the setting parameter from the NG image of the inspection process a graph generating unit for generating a graph in which an NG image measurement value, which is the value of the measurement parameter calculated according to, and the current threshold value are drawn;
Output unit for outputting the graph to a display device
A parameter adjustment support device characterized in that it comprises a. According to claim 1,
The parameter adjustment support device characterized in that the OK image measurement value, the NG image measurement value, and the threshold value are drawn on the graph in a manner in which at least one of a color or a figure is different. According to claim 1,
The graph generating unit determines whether adjustment of the parameter of the inspection process is necessary based on a relationship between the OK image measurement value, the NG image measurement value, and the threshold value in the inspection process for the adjustment target, and A parameter adjustment support device characterized in that the determination result is drawn on the graph. According to claim 3,
Parameter adjustment support device according to claim 1 , wherein the graph generating unit determines which adjustment of the threshold value and the setting parameter is required, and draws a result of the determination on the graph. According to claim 4,
The graph generating unit determines at least three states of the parameters need not be adjusted, the threshold needs to be adjusted, and the setting parameter needs to be adjusted.
The parameter adjustment support device, characterized in that the three states are drawn on the graph in different colors. According to any one of claims 1 to 5,
The parameter adjustment support device according to claim 1, wherein the output unit outputs an adjustment window for performing the parameter adjustment operation to the display device. According to claim 6,
2. The parameter adjustment support device according to claim 1, wherein the adjustment window displays a scatter diagram in which OK image measurement values of each OK image, NG image measurement values of each NG image, and the threshold are plotted. According to claim 7,
The adjustment window sets the OK image measurement value and the NG image measurement value drawn on the scatter plot to the OK image measurement value of an OK image outside the OK range defined by the threshold value and the NG image of an NG image included in the OK range. A parameter adjustment support device characterized in that switching is possible only with image measurement values. A parameter adjustment support method for supporting adjustment of parameters used in each inspection process for an image sensor capable of executing a plurality of inspection processes on an image,
The inspection process calculates a value of a predetermined measurement parameter from an image, compares the value of the measurement parameter with a threshold value, and outputs an OK or NG determination result;
The parameter includes a setting parameter that defines conditions for calculating the value of the measurement parameter and the threshold value,
A step of setting inspection processing of N items (N is an integer of 2 or more) to be adjusted among the plurality of inspection processing;
acquiring an OK image for which an OK determination is to be obtained in all of the N items of inspection processing and an NG image for which an NG judgment is to be obtained in at least one of the N items of inspection processing;
For each of the N inspection processes, an OK image measurement value, which is a value of the measurement parameter calculated from an OK image according to the current condition of the setting parameter, and a current condition of the setting parameter from the NG image of the inspection process a step of generating a graph in which an NG image measurement value, which is the value of the measurement parameter calculated according to , and the current threshold value are drawn;
Step of outputting the graph to a display device
Parameter adjustment support method comprising a. A recording medium on which a program for causing a processor to execute each step of the parameter adjustment support method according to claim 9 is non-temporarily recorded.

Images