Mold surface rust detection method and system

Abstract

The invention relates to a mold surface rust detection method and system, and relates to the technical field of mold defect detection.The method comprises the steps that a mold surface image is obtained; determining a suspected rust area on the mold surface image; determining a pixel depth value of each pixel point in the suspected rust region; determining adjacent points according to each pixel point, and determining an adjacent depth difference value according to the pixel depth values of the pixel points and the pixel depth values of the adjacent points; determining an adjacent deviation mean value according to all adjacent depth difference values determined by the single pixel point; calculating according to all adjacent deviation mean values to determine a flatness degree value; judging whether the flatness degree value is greater than a demand degree value; if yes, defining the current corresponding suspected rust area as an external influence area; and if not, defining the current corresponding suspected rust area as an actual rust area. The method has the effect of improving the reliability of the overall detection result of mold surface rust detection.

Description

Technical Field

[0001] The present application relates to the field of mold defect detection technology, and in particular to a mold surface rust detection method and system. Background Art

[0002] During mold manufacturing and maintenance, surface rust can seriously affect product quality and service life, so accurate detection and efficient removal of rust are crucial.

[0003] Currently, visual recognition technology is widely used in the industry for rust detection. This method, based on image processing or deep learning algorithms, identifies rusted areas by analyzing surface features such as color and texture. However, because surface features such as machining marks and oil residue visually resemble rust (e.g., appearing dark or with varying textures), traditional visual algorithms struggle to accurately distinguish them, resulting in a high number of false positives and low reliability of overall detection results. This leaves room for improvement. Summary of the Invention

[0004] In order to improve the reliability of the overall detection results of mold surface rust detection, the present application provides a mold surface rust detection method and system.

[0005] In a first aspect, the present application provides a method for detecting rust on a mold surface, which adopts the following technical solution: A method for detecting rust on a mold surface, comprising: Acquire mold surface images; Perform feature recognition on the mold surface image to identify suspected rust areas; Determine the pixel depth value of each pixel in the suspected rust area based on the mold surface image; Determine adjacent points based on each pixel in the suspected rust area, and perform difference calculation based on the pixel depth value of the pixel point and the pixel depth value of the adjacent point to determine the adjacent depth difference; Calculate the adjacent depth difference values ​​determined by a single pixel to determine the adjacent deviation mean; The flatness value is determined by calculating the mean of all adjacent deviations; Determine whether the flatness value is greater than the preset required degree value; If the flatness value is greater than the required value, the corresponding suspected rust area is defined as the external impact area; If the flatness value is not greater than the required flatness value, the corresponding suspected rust area is defined as the actual rust area.

[0006] Optionally, the step of performing feature recognition on the mold surface image to determine the suspected rust area includes: Obtaining the pixel chromaticity value of each pixel on the mold surface image; Divide each pixel into a preset monomer color set according to the pixel chromaticity value, where each monomer color set corresponds to a color chromaticity range; Count the pixels within each monomer color set to determine the number within the set; According to the preset sorting rules, the number within the set with the largest value is determined, and the color chromaticity range corresponding to the monomer color set corresponding to the number within the set is defined as the main chromaticity range; Determine the mold metal type corresponding to the main body chromaticity range based on the preset type matching relationship; The corresponding preset feature recognition database is called according to the mold metal type, and the mold surface image is input into the feature recognition database to determine the suspected rust area.

[0007] Optionally, after the mold metal type is determined, the mold surface rust detection method also includes: Counting by die metal type to determine the number of metal types; Determine whether the number of metal types is one; If the number of metal types is one, then the corresponding feature recognition database is determined according to the current mold metal type to perform feature recognition; If the number of metal types is not one, the corresponding mold metal type is defined as an alternative metal type, and the single color set with a number greater than a preset reference effective number within the set except the single color set corresponding to the main body chromaticity range is defined as a rust color set; Calculate the pixel ratio of each rust color set based on the number of pixels in each rust color set; Determine the standard pixel ratio of the candidate metal type under the rust color set according to a preset ratio matching relationship; The similarity value is determined by calculating the ratio of all set pixels and the corresponding standard pixel ratio, and the corresponding feature recognition database is determined according to the candidate metal type corresponding to the largest similarity value for feature recognition.

[0008] Optionally, the step of determining the standard pixel ratio of the candidate metal type in the rust color set according to a preset ratio matching relationship includes: Construct a detection interval on the preset time axis with the current time point as the end point and a width of the preset detection duration; Obtain external environmental parameters at each time point during the detection interval; Analyze all external environmental parameters to determine the comprehensive external environment; The corresponding ratio matching relationship is called according to the comprehensive external environment, and the standard pixel ratio is determined according to the ratio matching relationship.

[0009] Optionally, the step of calculating the flatness value based on the mean of all adjacent deviations includes: Randomly select one adjacent deviation mean from all adjacent deviation means as the key deviation mean, and define the remaining adjacent deviation means as secondary deviation means; The difference between the key deviation mean and the minor deviation mean is calculated to determine the single effective deviation; The overall effective deviation is determined by summing up all the individual effective deviations, and the emphasis level value is determined by calculating based on the overall effective deviation and the preset fixed calculation parameters; The maximum emphasis value is determined according to the sorting rule, and the emphasis value is determined as the flatness value.

[0010] Optionally, after the actual rust area is determined, the mold surface rust detection method further includes: Determine the actual area based on each actual rust area; The upper limit area corresponding to the comprehensive external environment and the detection time is determined based on the preset upper limit matching relationship; Determine whether the actual area is greater than the upper limit area; If the actual area is not greater than the upper limit area, the currently determined actual rust area is maintained; If the actual area is larger than the upper limit area, the current actual rust area is defined as the waiting review area.

[0011] Optionally, after the waiting review area is determined, the mold surface rust detection method also includes: Determine the distance between regions based on the area waiting for review and each actual rust area, and define the region distance with the smallest value as the closest distance; Determine the close distance range based on the preset close distance and the closest distance; Count the actual rust areas within a close distance range to determine the number within the range; The reliability parameters are determined based on the closest distance, the number of items within the range, and the preset reliability calculation parameters.

[0012] Optionally, after the actual rust area is determined, the mold surface rust detection method further includes: Randomly determine the entry point and exit point on the contour line of each actual rust area, and randomly sort the actual rust areas in order to determine the area processing order; Determine the rust removal movement path according to the area processing sequence and the entry point and exit point on the actual rust area, and determine the rust removal movement distance according to the rust removal movement path; The waiting review area with a reliability parameter greater than a preset benchmark parameter is defined as a priority area, and the priority areas passed through on the rust removal moving path are counted to determine the number of priority areas; Determine the movement correction distance corresponding to the priority number according to the preset correction matching relationship, and update the rust removal movement distance according to the movement correction distance; The rust removal moving distance with the smallest value is determined according to the sorting rule, and the rust removal moving path corresponding to the rust removal moving distance is output as a rust removal guidance path.

[0013] In a second aspect, the present application provides a mold surface rust detection system, which adopts the following technical solutions: A mold surface rust detection system, comprising: An acquisition module, used for acquiring a mold surface image; A processing module, connected to the acquisition module and the judgment module, for storing and processing information; The judgment module is connected with the acquisition module and the processing module and is used for judging the information; The processing module performs feature recognition on the mold surface image to determine suspected rust areas; The processing module determines the pixel depth value of each pixel point in the suspected rust area based on the mold surface image; The processing module determines adjacent points based on each pixel in the suspected rust area, and performs difference calculation based on the pixel depth value of the pixel point and the pixel depth value of the adjacent point to determine the adjacent depth difference; The processing module calculates all adjacent depth differences determined by a single pixel point to determine an adjacent deviation mean; The processing module calculates the flatness value based on the average of all adjacent deviations; The judging module judges whether the flatness value is greater than a preset required degree value; If the judging module determines that the flatness value is greater than the required value, the processing module defines the corresponding suspected rust area as an external impact area; If the judging module determines that the flatness value is not greater than the required flatness value, the processing module defines the corresponding suspected rust area as an actual rust area.

[0014] In summary, this application includes at least one of the following beneficial technical effects: When inspecting rust on the mold surface, the system not only uses visual feature analysis but also analyzes the concave and convex conditions within the suspected area to determine whether it is rust, thereby improving the accuracy of mold surface rust detection. The mold metal material type is identified through images, and the appropriate feature recognition database can be used to inspect the mold surface image, thereby improving the accuracy of rust detection. During rust detection, some unreasonable rust areas are marked to facilitate staff intervention. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 It is a flow chart of the method for detecting rust on the mold surface.

[0016] Figure 2 It is a module flow chart of the mold surface rust detection method. DETAILED DESCRIPTION

[0017] In order to make the purpose, technical solutions and advantages of this application more clear, the following Figure 1-Figure 2 It should be understood that the specific embodiments described herein are only used to explain the present application and are not intended to limit the present application.

[0018] The embodiments of the present application are described in further detail below with reference to the accompanying drawings.

[0019] The present application discloses a method for detecting rust on a mold surface. Figure 1 The method flow of the mold surface rust detection method includes the following steps: Step S100: Acquire a mold surface image.

[0020] The mold surface image is an image of the surface of the mold that needs to be inspected for rust, obtained by an image capturing device.

[0021] Step S101: performing feature recognition on the mold surface image to determine suspected rust areas.

[0022] Suspected rust areas are areas that can be determined to contain rust through feature analysis. This feature recognition method uses a neural network to learn various samples beforehand to construct a corresponding recognition database. The mold surface image is then input into the corresponding recognition database for recognition processing. This method is a conventional technical means for mold surface rust recognition based on visual recognition and will not be described in detail here.

[0023] Step S102 : determining the pixel depth value of each pixel in the suspected rust area according to the mold surface image.

[0024] The pixel depth value is the depth of the position corresponding to each pixel point, which can effectively reflect the concave and convex conditions of the actual position corresponding to the pixel point, and the mold surface image can be obtained through the binocular camera.

[0025] Step S103 : determining adjacent points according to each pixel in the suspected rust area, and performing difference calculation based on the pixel depth value of the pixel and the pixel depth value of the adjacent point to determine the adjacent depth difference.

[0026] The adjacent points are the remaining pixel points adjacent to a single pixel point in the suspected rust area. Adjacent means that there is a common edge between the two pixel points. The adjacent depth difference is the difference in pixel depth values ​​between the two pixel points, which is an absolute value.

[0027] Step S104: Calculate all adjacent depth differences determined for a single pixel to determine an adjacent deviation mean.

[0028] The mean of adjacent deviations is the average of all adjacent depth differences determined by a single pixel.

[0029] Step S105: Calculate the flatness value based on the average of all adjacent deviations.

[0030] The flatness value reflects the degree of change in the position corresponding to each pixel point in the suspected defect area. The larger the value, the smoother the concave-convex change. It can be determined by calculating the average value of each adjacent deviation mean and then taking the inverse of the average value. It can also be determined by referring to the method of steps S500-S503. The specific setting is made by the staff according to the actual situation.

[0031] Step S106: Determine whether the flatness value is greater than a preset required flatness value.

[0032] The required degree value is the minimum flatness value set by the staff to determine that the changes in each pixel point in the suspected rust area are relatively stable. The purpose of the judgment is to know whether the unevenness of each pixel point is smooth. Since the unevenness changes in rust are random, while the unevenness changes in machining lines, oil residues, etc. are relatively stable, the flatness value is judged to determine the area where rust actually exists.

[0033] Step S1061: If the flatness value is greater than the required flatness value, the corresponding suspected rust area is defined as an external impact area.

[0034] When the flatness value is greater than the required value, it indicates that the current suspected rust area is more likely to be misdetected, so it can be defined as an external impact area.

[0035] Step S1062: If the flatness value is not greater than the required flatness value, the corresponding suspected rust area is defined as an actual rust area.

[0036] When the flatness value is not greater than the required value, it indicates that there is a random change in the concavity and convexity, that is, the current corresponding area is likely to have rust. Therefore, it is defined as the actual rust area for identification, thereby improving the accuracy of rust detection.

[0037] The steps of performing feature recognition on the mold surface image to identify suspected rust areas include: Step S200: Obtaining the pixel chromaticity value of each pixel on the mold surface image.

[0038] The pixel chromaticity value is the chromaticity parameter of each pixel, represented by the three primary colors of red, yellow and green.

[0039] Step S201: Divide each pixel into a preset single color set according to the pixel chromaticity value, wherein each single color set corresponds to a color chromaticity range.

[0040] The single color set is a set of corresponding colors set by the staff, such as silver, brown, etc. Each color corresponds to a color chromaticity range. The pixels are divided according to which color chromaticity range the pixel chromaticity value falls in, so that pixels with similar colors can be summarized for subsequent analysis.

[0041] Step S202: Count the pixels in each monomer color set to determine the number of pixels in the set.

[0042] The internal number of the set is the total number of pixels in the monomer color set, which can be determined by counting each pixel one by one.

[0043] Step S203: determining the internal quantity of the set with the largest value according to a preset sorting rule, and defining the color chromaticity range corresponding to the monomer color set corresponding to the internal quantity of the set as the main chromaticity range.

[0044] The sorting rule is a method set by the staff to sort the size of the values, such as the bubble method. The sorting rule can be used to determine the number within the set with the largest value. Since the color that appears most frequently in the mold must be the color corresponding to the mold itself, the color chromaticity range corresponding to the monomer color set corresponding to the number within the set is the chromaticity range of the mold metal. At this time, it is defined as the main chromaticity range for identification to facilitate subsequent analysis.

[0045] Step S204: Determine the mold metal type corresponding to the main body chromaticity range according to a preset type matching relationship.

[0046] The mold metal type is the metal type of the current mold identified by color. Different main body chromaticity ranges indicate different colors, and the corresponding mold metal types are also different. The type matching relationship between the two can be determined and entered in advance by the staff. For example, Cr12MoV mold steel is dark blue, while SUS304 stainless steel is silvery white.

[0047] Step S205 : calling a corresponding preset feature recognition database according to the metal type of the mold, and inputting the mold surface image into the feature recognition database to determine the suspected rust area.

[0048] Due to the different components of molds of different metal types, the color and texture of the rust generated will also be different. At this time, calling out the appropriate feature recognition database based on the current mold metal type can better perform feature recognition on the image. Each mold metal type corresponds to a feature recognition database, which is obtained by staff through prior training and entered into storage for call.

[0049] After the mold metal type is determined, the mold surface rust detection method also includes: Step S300: Counting the metal types of the mold to determine the number of metal types.

[0050] The number of metal types is the number of determined mold metal types.

[0051] Step S301: Determine whether the number of metal types is one.

[0052] The purpose of the judgment is to know whether the specific metal type of the mold can be determined.

[0053] Step S3011: If the number of metal types is one, then determine the corresponding feature recognition database according to the current mold metal type to perform feature recognition.

[0054] When the number of metal types is one, it means that there is only one mold metal type that meets the requirements. At this time, the corresponding feature recognition database can be called according to the mold metal type.

[0055] Step S3012: If the number of metal types is not one, the corresponding mold metal type is defined as an alternative metal type, and the monomer color set with a number greater than a preset benchmark effective number within the set except the monomer color set corresponding to the main body chromaticity range is defined as a rust color set.

[0056] When the number of metal types is not one, it means that there are multiple mold metal types that meet the requirements. Therefore, alternative metal types are defined to distinguish different mold metal types for subsequent analysis. The benchmark effective number is the minimum number within the set set by the staff that is required for the representative color to be present. By defining the rust color set, the current rust color that will appear is identified for subsequent analysis.

[0057] Step S302: performing calculations based on the number of pixels within each rust color set to determine the pixel ratio of each rust color set.

[0058] The set pixel ratio is the ratio of pixels of a single color to pixels in all rust color sets, and is determined by dividing the number of pixels within the set in a single rust color set by the sum of the number of pixels within the set in all rust color sets.

[0059] Step S303: determining a standard pixel ratio of the candidate metal type under the rust color set according to a preset ratio matching relationship.

[0060] The standard pixel ratio is the theoretical ratio that the pixels corresponding to the colors of each rust color set should have when rust occurs on a mold of the alternative metal type. The matching relationship between the three ratios is determined by the staff through multiple tests in advance and will not be elaborated here.

[0061] Step S304: Calculate the similarity value based on all the set pixel ratios and the corresponding standard pixel ratios, and determine the corresponding feature recognition database based on the candidate metal type corresponding to the largest similarity value for feature recognition.

[0062] The similarity value reflects the similarity between the current rust color distribution and the rust color distribution of the corresponding type of metal under standard conditions. The larger the value, the more similar the two are. The similarity value is determined by calculating the absolute difference between the set pixel ratio of each rust color set and the corresponding standard pixel ratio, and then adding all the differences to obtain the inverse. The alternative metal type corresponding to the largest similarity value is the metal type that best matches the current rust distribution. Therefore, the corresponding feature recognition database can be called to identify the image based on the alternative metal type.

[0063] The steps of determining the standard pixel ratio of the candidate metal type under the rust color set according to the preset ratio matching relationship include: Step S400: constructing a detection interval on a preset time axis with the current time point as the end point and a width of a preset detection duration.

[0064] The time axis is a coordinate axis formed by the combination of various time points. The coordinate axis points from the time points that have passed to the time points that have not yet arrived, where the time points that have passed are on the left side of the time axis, and the left side is defined as the front end of the time axis; the detection duration is the interval between the current time point and the time point when the previous rust removal was completed. By constructing the detection interval, the data after the previous rust removal is completed can be obtained and analyzed.

[0065] Step S401: Obtain external environmental parameters at each time point in the detection interval.

[0066] The external environmental parameters are the environmental parameters of the environment where the mold is located at each time point in the detection range, such as temperature, humidity, and other parameters that may affect the rusting condition, where each time point is counted in days.

[0067] Step S402: Analyze all external environment parameters to determine the comprehensive external environment.

[0068] The comprehensive external environment reflects the comprehensive environmental conditions that the mold has experienced since the last rust removal was completed. For example, a high temperature and humid environment can be determined by collectively analyzing various external environmental parameters. The specific analysis method can be determined in advance by the staff and will not be elaborated here.

[0069] Step S403: calling a corresponding ratio matching relationship according to the comprehensive external environment, and determining a standard pixel ratio according to the ratio matching relationship.

[0070] The proportion of rust produced under different comprehensive external environments is different, so the corresponding proportion matching relationship is also different. At this time, by calling the appropriate proportion matching relationship according to the actual external environment, a more accurate standard pixel proportion can be determined, so that the metal type of the mold can be better determined.

[0071] The steps for calculating the flatness value based on the mean of all adjacent deviations include: Step S500 : randomly selecting one adjacent deviation mean from all adjacent deviation means as a key deviation mean, and defining the remaining adjacent deviation means as secondary deviation means.

[0072] By defining the key deviation mean and the secondary deviation mean, we can distinguish different adjacent deviation means to facilitate subsequent analysis.

[0073] Step S501: performing difference calculation based on the mean of the key deviation and the mean of each secondary deviation to determine the single effective deviation.

[0074] The single effective deviation is the difference between the mean of the key deviation and the mean of each minor deviation, and the difference is the absolute value.

[0075] Step S502: performing a sum calculation based on all individual effective deviations to determine the overall effective deviation, and performing a calculation based on the overall effective deviation and preset fixed calculation parameters to determine the emphasis level value.

[0076] The overall effective deviation is the sum of all individual effective deviations, and the fixed calculation parameter is a fixed value used for calculation. The emphasis level value can be determined by dividing the fixed calculation parameter by the overall effective deviation.

[0077] Step S503: determining the maximum emphasis value according to the sorting rule, and determining the emphasis value as the flatness value.

[0078] The sorting rules can be used to determine the most important value, which is the value that best represents the concave and convex conditions of the area. In this case, it can be determined as the flatness value.

[0079] After the actual rust area is determined, the mold surface rust detection method also includes: Step S600: determining the actual area of ​​each rust area.

[0080] The actual area is the area of ​​a single actual rust area.

[0081] Step S601: determining the upper limit area corresponding to the comprehensive external environment and the detection duration according to a preset upper limit matching relationship.

[0082] The upper limit area is the maximum area of ​​a single rust area that will appear after the mold is tested under a comprehensive external environment under theoretical conditions. The upper limit matching relationship between the three can be determined by the staff through multiple tests in advance, which will not be elaborated here.

[0083] Step S602: Determine whether the actual area is greater than the upper limit area.

[0084] The purpose of the judgment is to determine whether the actual rust area currently determined is reasonable, that is, to determine whether the actual rust area determined is reliable.

[0085] Step S6021: If the actual area is not greater than the upper limit area, the currently determined actual rust area is maintained.

[0086] When the actual area is not greater than the upper limit area, it means that the actual rust area currently determined is more reliable, so it can be maintained.

[0087] Step S6022: If the actual area is greater than the upper limit area, the current actual rust area is defined as a waiting review area.

[0088] When the actual area is larger than the upper limit area, it means that the actual rust area currently determined has some reliability issues. At this time, it is defined as a waiting review area to facilitate staff intervention and analysis.

[0089] After the area to be reviewed is determined, the mold surface rust detection method also includes: Step S700: determining the distance between regions based on the region waiting for review and each actual rust region, and defining the region distance with the smallest value as the closest distance.

[0090] The regional separation distance is the distance between the waiting-for-review area and the actual rust area. It is the minimum straight-line distance between a point in the waiting-for-review area and a point in the actual rust area. The closest separation distance is defined to distinguish the regional separation distances of the closest actual rust area for subsequent analysis.

[0091] Step S701: determining a close distance range according to a preset close distance and a closest distance.

[0092] The close distance is a fixed distance set by the staff, and the close distance range is a range with the closest distance as the lower endpoint and the sum of the closest distance and the close distance as the upper endpoint.

[0093] Step S702: Counting actual rust areas within a similar distance range to determine the number within the range.

[0094] The number within the range is the total number of actual rust areas within a similar distance range, that is, it reflects the total number of actual rust areas surrounding the area currently waiting for review.

[0095] Step S703: performing calculation according to the closest distance, the number within the range, and preset reliability calculation parameters to determine the reliability parameter.

[0096] The reliability parameter reflects the possibility that the current waiting-for-review area can still be the actual rust area. The larger the value, the higher the possibility that the waiting-for-review area is still the actual rust area after the review. The calculation formula is: ,in is the reliability parameter, is the number within the range, is the closest distance, as well as It is a reliable calculation parameter and is a fixed value.

[0097] After the actual rust area is determined, the mold surface rust detection method also includes: Step S800: randomly determining an entry point and an exit point on the contour line of each actual rust area, and randomly sorting each actual rust area to determine the area processing order.

[0098] The entry point and the exit point are both random positions on the contour line of the actual rust area, and the area processing order is the order obtained by randomly sorting the actual rust areas.

[0099] Step S801: determining a rust removal moving path according to the region processing sequence and the entry point and exit point on the actual rust region, and determining a rust removal moving distance according to the rust removal moving path.

[0100] The rust removal movement path is the operation path of the laser rust removal device that can perform rust removal on the actual rust area in sequence according to the area processing order, wherein the device enters the actual rust area from the entry point and moves out from the exit point, and moves from the current exit point to the entry point of the next actual rust area in the area processing order, repeating the above operation until the rust removal operation of all actual rust areas is completed. At this time, the path that the device needs to move outside the actual rust area is the rust removal movement path; the rust removal movement distance is the distance that the device needs to move on the rust removal movement path.

[0101] Step S802: defining the waiting review area whose reliability parameter is greater than the preset reference parameter as a priority area, and counting the priority areas passed through on the rust removal moving path to determine the number of priority areas.

[0102] The baseline parameter is the minimum reliability parameter required when the area awaiting review has a high probability of being defined as an actual rust area after subsequent review. Priority areas are defined to distinguish different areas awaiting review for the convenience of subsequent analysis; the priority number is the number of priority areas on the rust removal movement path.

[0103] Step S803: determining a movement correction distance corresponding to the priority number according to a preset correction matching relationship, and updating the rust removal movement distance according to the movement correction distance.

[0104] The moving correction distance is the distance value used to correct and update the rust removal moving distance. Different priority numbers indicate that the number of areas that the current rust removal device can process when moving along the rust removal moving path is different. Therefore, the value of each path is also different. At this time, the corresponding moving correction distance is also different. The correction matching relationship between the two is determined by the staff through multiple tests in advance. It is necessary to ensure that the larger the priority number, the larger the corresponding moving correction distance; the correction and update method for the rust removal moving distance is to subtract the moving correction distance from the rust removal moving distance.

[0105] Step S804: determining the rust removal moving distance with the smallest value according to the sorting rule, and outputting the rust removal moving path corresponding to the rust removal moving distance as a rust removal guiding path.

[0106] The sorting rules can be used to determine the rust removal moving distance with the minimum value, that is, the current corresponding rust removal moving path has the shortest moving distance or the highest value while meeting the rust removal requirements. Therefore, it is defined as the rust removal guidance path and output for subsequent rust removal reference.

[0107] Reference Figure 2 Based on the same inventive concept, an embodiment of the present invention provides a mold surface rust detection system, comprising: An acquisition module, used for acquiring a mold surface image; A processing module, connected to the acquisition module and the judgment module, for storing and processing information; The judgment module is connected with the acquisition module and the processing module and is used for judging the information; The processing module performs feature recognition on the mold surface image to determine suspected rust areas; The processing module determines the pixel depth value of each pixel point in the suspected rust area based on the mold surface image; The processing module determines adjacent points based on each pixel in the suspected rust area, and performs difference calculation based on the pixel depth value of the pixel point and the pixel depth value of the adjacent point to determine the adjacent depth difference; The processing module calculates all adjacent depth differences determined by a single pixel point to determine an adjacent deviation mean; The processing module calculates the flatness value based on the average of all adjacent deviations; The judging module judges whether the flatness value is greater than a preset required degree value; If the judging module determines that the flatness value is greater than the required value, the processing module defines the corresponding suspected rust area as an external impact area; If the judging module determines that the flatness value is not greater than the required value, the processing module defines the corresponding suspected rust area as the actual rust area; A feature recognition database determination module is used to determine a suitable feature recognition database to perform feature recognition on an image; Alternative metal type screening module, which screens the alternative metal types to determine the appropriate feature recognition database; A standard pixel ratio determination module is used to determine an appropriate standard pixel ratio for use; A flatness value determination module is used to determine a relatively accurate flatness value to identify the rust condition; A review situation determination module is used to identify and determine unreasonable rust areas that need to be reviewed; The review situation prediction module is used to predict the possible results of the area waiting for review; The rust removal path guidance module is used to output a suitable rust removal path to guide subsequent rust removal operations.

[0108] Those skilled in the art will clearly understand that for the sake of convenience and brevity, the division of the above-mentioned functional modules is only used as an example for illustration. In actual applications, the above-mentioned functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above. The specific working processes of the above-mentioned systems, devices, and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

Claims

1. A method for detecting rust on a mold surface, characterized in that: include: Acquire mold surface images; Perform feature recognition on the mold surface image to identify suspected rust areas; Determine the pixel depth value of each pixel in the suspected rust area based on the mold surface image; Determine adjacent points based on each pixel in the suspected rust area, and perform difference calculation based on the pixel depth value of the pixel point and the pixel depth value of the adjacent point to determine the adjacent depth difference; Calculate the adjacent depth difference values ​​determined by a single pixel to determine the adjacent deviation mean; The flatness value is determined by calculating the mean of all adjacent deviations; Determine whether the flatness value is greater than the preset required degree value; If the flatness value is greater than the required value, the corresponding suspected rust area is defined as the external impact area; If the flatness value is not greater than the required flatness value, the corresponding suspected rust area is defined as the actual rust area.

2. The method for detecting rust on a mold surface according to claim 1, wherein: The steps of performing feature recognition on the mold surface image to identify suspected rust areas include: Obtaining the pixel chromaticity value of each pixel on the mold surface image; Divide each pixel into a preset monomer color set according to the pixel chromaticity value, where each monomer color set corresponds to a color chromaticity range; Count the pixels within each monomer color set to determine the number within the set; According to the preset sorting rules, the number within the set with the largest value is determined, and the color chromaticity range corresponding to the monomer color set corresponding to the number within the set is defined as the main chromaticity range; Determine the mold metal type corresponding to the main body chromaticity range based on the preset type matching relationship; The corresponding preset feature recognition database is called according to the mold metal type, and the mold surface image is input into the feature recognition database to determine the suspected rust area.

3. The method for detecting rust on a mold surface according to claim 2, wherein: After the mold metal type is determined, the mold surface rust detection method also includes: Counting by die metal type to determine the number of metal types; Determine whether the number of metal types is one; If the number of metal types is one, then the corresponding feature recognition database is determined according to the current mold metal type to perform feature recognition; If the number of metal types is not one, the corresponding mold metal type is defined as an alternative metal type, and the single color set with a number greater than a preset reference effective number within the set except the single color set corresponding to the main body chromaticity range is defined as a rust color set; Calculate the pixel ratio of each rust color set based on the number of pixels in each rust color set; Determine the standard pixel ratio of the candidate metal type under the rust color set according to a preset ratio matching relationship; The similarity value is determined by calculating the ratio of all set pixels and the corresponding standard pixel ratio, and the corresponding feature recognition database is determined according to the candidate metal type corresponding to the largest similarity value for feature recognition.

4. The method for detecting rust on a mold surface according to claim 3, wherein: The steps of determining the standard pixel ratio of the candidate metal type under the rust color set according to the preset ratio matching relationship include: Construct a detection interval on the preset time axis with the current time point as the end point and a width of the preset detection duration; Obtain external environmental parameters at each time point during the detection interval; Analyze all external environmental parameters to determine the comprehensive external environment; The corresponding ratio matching relationship is called according to the comprehensive external environment, and the standard pixel ratio is determined according to the ratio matching relationship.

5. The method for detecting rust on a mold surface according to claim 1, wherein: The steps for calculating the flatness value based on the mean of all adjacent deviations include: Randomly select one adjacent deviation mean from all adjacent deviation means as the key deviation mean, and define the remaining adjacent deviation means as secondary deviation means; The difference between the key deviation mean and the minor deviation mean is calculated to determine the single effective deviation; The overall effective deviation is determined by summing up all the individual effective deviations, and the emphasis level value is determined by calculating based on the overall effective deviation and the preset fixed calculation parameters; The maximum emphasis value is determined according to the sorting rule, and the emphasis value is determined as the flatness value.

6. The method for detecting rust on a mold surface according to claim 4, wherein: After the actual rust area is determined, the mold surface rust detection method also includes: Determine the actual area based on each actual rust area; The upper limit area corresponding to the comprehensive external environment and the detection time is determined based on the preset upper limit matching relationship; Determine whether the actual area is greater than the upper limit area; If the actual area is not greater than the upper limit area, the currently determined actual rust area is maintained; If the actual area is larger than the upper limit area, the current actual rust area is defined as the waiting review area.

7. The method for detecting rust on a mold surface according to claim 6, wherein: After the area to be reviewed is determined, the mold surface rust detection method also includes: Determine the distance between regions based on the area waiting for review and each actual rust area, and define the region distance with the smallest value as the closest distance; Determine the close distance range based on the preset close distance and the closest distance; Count the actual rust areas within a close distance range to determine the number within the range; The reliability parameters are determined based on the closest distance, the number of items within the range, and the preset reliability calculation parameters.

8. The method for detecting rust on a mold surface according to claim 7, wherein: After the actual rust area is determined, the mold surface rust detection method also includes: Randomly determine the entry point and exit point on the contour line of each actual rust area, and randomly sort the actual rust areas in order to determine the area processing order; Determine the rust removal movement path according to the area processing sequence and the entry point and exit point on the actual rust area, and determine the rust removal movement distance according to the rust removal movement path; The waiting review area with a reliability parameter greater than a preset benchmark parameter is defined as a priority area, and the priority areas passed through on the rust removal moving path are counted to determine the number of priority areas; Determine the movement correction distance corresponding to the priority number according to the preset correction matching relationship, and update the rust removal movement distance according to the movement correction distance; The rust removal moving distance with the smallest value is determined according to the sorting rule, and the rust removal moving path corresponding to the rust removal moving distance is output as a rust removal guidance path.

9. A mold surface rust detection system, characterized in that: include: An acquisition module, used for acquiring a mold surface image; A processing module, connected to the acquisition module and the judgment module, for storing and processing information; The judgment module is connected with the acquisition module and the processing module and is used for judging the information; The processing module performs feature recognition on the mold surface image to determine suspected rust areas; The processing module determines the pixel depth value of each pixel point in the suspected rust area based on the mold surface image; The processing module determines adjacent points based on each pixel in the suspected rust area, and performs difference calculation based on the pixel depth value of the pixel point and the pixel depth value of the adjacent point to determine the adjacent depth difference; The processing module calculates all adjacent depth differences determined by a single pixel point to determine an adjacent deviation mean; The processing module calculates the flatness value based on the average of all adjacent deviations; The judging module judges whether the flatness value is greater than a preset required degree value; If the judging module determines that the flatness value is greater than the required value, the processing module defines the corresponding suspected rust area as an external impact area; If the judging module determines that the flatness value is not greater than the required flatness value, the processing module defines the corresponding suspected rust area as an actual rust area.

Images