Monitoring and early warning method and system for luggage production and computing equipment

Abstract

The invention discloses a monitoring and early warning method and system for luggage production and computing equipment, relates to the technical field of luggage production defect detection, and solves the problems that in the prior art, it is not considered that a reflective material in a luggage presents pure white in a gray image and is mixed with the white of the luggage, so that real holes or stitches are covered, and the quality of the luggage is influenced. And the accuracy of the detection result is reduced. Normal detection image sets of different types of bags are obtained through pre-detection; gray level images of different types of bags on the production line are collected in real time through a visual sensor; judging whether a light reflection phenomenon exists in the gray images of the corresponding types of bags on the same production line or not; judging whether a visual sensor needs to be added to the corresponding production line; determining the number of added visual sensors of the corresponding production line according to the reflection condition; a defect detection model is constructed to carry out fault detection on the gray level image with the light reflection phenomenon to obtain a fault type, and an early warning signal is sent out; according to the invention, the accuracy of luggage production detection is improved.

Description

Technical Field

[0001] The present invention belongs to the field of bag production defect detection, relates to equipment detection technology, and specifically provides a monitoring and early warning method, system and computing equipment for bag production. Background Art

[0002] Under the existing technical background, the bags in the production line are monitored through hardware such as industrial-grade vision systems and high-sampling rate sensors. This method is costly and difficult for small and medium-sized enterprises to afford. At the same time, all production lines in the factory are basically monitored uniformly. If sudden failures are not discovered in time, inevitable economic losses will occur. By considering factors such as the material and price of bags, the production lines are classified and monitored, thereby reducing monitoring costs and improving monitoring efficiency.

[0003] The prior art (the invention patent application with the publication number CN118837360A) discloses a method and system for detecting production defects of multifunctional luggage, which belongs to the technical field of production defect detection of multifunctional luggage. The method comprises: establishing a display model of multifunctional luggage; performing a benchmark test on the produced multifunctional luggage to obtain a corresponding benchmark test result; performing a sewing test on the multifunctional luggage to determine the sewing area of ​​the multifunctional luggage; collecting a sewing image of the sewing area; performing grayscale processing on the sewing image to obtain a sewing grayscale image; obtaining a corresponding sewing background image in the sewing image; performing grayscale adjustment on the sewing grayscale image according to the sewing background image; determining each hole area and stitching area through the adjusted sewing grayscale image; analyzing each hole area and stitching area respectively to obtain a hole detection result and a stitching detection result, and integrating the hole detection result and the stitching detection result into a sewing detection result; marking the sewing detection result and the benchmark detection result correspondingly in the display model;

[0004] The prior art obtains hole detection results and stitch detection results by analyzing each hole area and stitch area separately, but does not take into account the presence of reflective materials in the luggage. The reflective area in the reflective material appears pure white in the grayscale image, which is confused with the white color of the luggage itself, thereby covering up the real holes or stitches, resulting in a technical problem of reduced accuracy of the detection results and increased possibility of errors;

[0005] The present invention provides a monitoring and early warning method, system and computing device for bag production to solve the above technical problems. Summary of the invention

[0006] The present invention aims to solve at least one of the technical problems existing in the prior art; to this end, the present invention proposes a monitoring and early warning method, system and computing device for luggage production, which are used to solve the technical problem that the prior art analyzes each hole area and stitching area separately, but does not take into account that the reflective area in the reflective material in the luggage appears pure white in the grayscale image, which is confused with the white color of the luggage itself, thereby covering up the real holes or stitching, resulting in a decrease in the accuracy of the detection results.

[0007] To achieve the above-mentioned purpose, the first aspect of the present invention provides a monitoring and early warning method for bag production, comprising:

[0008] S100: obtaining normal inspection image sets of different types of bags through pre-inspection; collecting grayscale images of different types of bags on the production line in real time through visual sensors;

[0009] S200: Determine whether there is a reflection phenomenon in the grayscale image of the corresponding type of bags on the same production line;

[0010] Yes, the reflectivity of the luggage is determined based on the reflectivity in the grayscale image;

[0011] If not, determine whether the reflective conditions of the corresponding types of bags on all production lines are obtained; if yes, mark as end of determination; if not, repeat step S200;

[0012] S300: Determine whether the corresponding production line needs to add a visual sensor;

[0013] If yes, determine the number of visual sensors to be added to the corresponding production line based on the reflection situation; determine whether the reflection situation after adding the visual sensors has degraded to a preset degree; if yes, stop adding the number of sensors and complete the production monitoring process; if no, send a reminder signal;

[0014] If not, proceed to step S400;

[0015] S400: Construct a defect detection model to perform fault detection on the grayscale image with reflection phenomenon to obtain the fault type and issue a warning signal.

[0016] Preferably, the step of obtaining a normal detection image set of different types of bags through pre-detection includes:

[0017] Obtain the type of bags to be detected and collect bag images under standard viewing angles; mark the bag images with preset white areas in the HSV color space; the standard viewing angle is the viewing angle of the bag images collected by the visual sensor in the absence of reflective light sources in the production line;

[0018] The luggage image under the standard viewing angle is preprocessed to obtain a pre-grayscale image, and the over-exposed grayscale value range corresponding to the preset white area of ​​the luggage itself in the pre-grayscale image is obtained; the pre-grayscale image within the over-exposed grayscale value range is used as the normal detection image set.

[0019] The present invention collects luggage images by setting a standard viewing angle, that is, without a reflective light source, in order to eliminate ambient light interference, ensure image feature consistency, and reduce grayscale value fluctuations caused by angle deviation; the saturation and brightness channels of HSV can effectively distinguish white from other color areas, avoiding mislabeling of other colors as white areas.

[0020] Preferably, the step of judging whether there is a reflection phenomenon in the grayscale images of bags of corresponding types on the same production line includes:

[0021] Retrieve a normal detection image set, and segment the normal overexposed area in the normal detection image set using the grayscale threshold segmentation method;

[0022] Retrieve grayscale images of different types of bags on the production line and count the overexposed areas within the overexposed grayscale range in the grayscale images;

[0023] Determine whether the area of ​​the overexposed region exceeds the corresponding normal overexposed region in the normal detection image set; if yes, mark the grayscale image of the corresponding type of luggage on the corresponding production line as having reflection; if no, mark the grayscale image of the corresponding type of luggage on the corresponding production line as having no reflection.

[0024] The present invention classifies bags on different production lines according to the reflective phenomenon, which is conducive to further improvement of bags containing reflective materials in a more targeted manner.

[0025] Preferably, determining the reflective condition of the luggage according to the reflective phenomenon in the grayscale image includes:

[0026] Count the number of all grayscale images marked as having reflections on the same production line; remove the normal overexposed area contained in the corresponding normal inspection image set from the overexposed area of ​​the grayscale image marked as having reflections to obtain the remaining reflection area; remove the normal overexposed area from the grayscale image to obtain the remaining area, and divide the remaining reflection area by the remaining area to obtain the remaining reflection area ratio;

[0027] Set a first threshold value S1 and a second threshold value S2, and mark the reflective situation of the corresponding bag in the interval (0, S1] as slight;

[0028] The reflective condition of the corresponding bags and bags whose remaining reflective area ratio is in the interval (S1, S2] is marked as moderate;

[0029] The reflective situation of the corresponding bags and suitcases whose remaining reflective area ratio is in the interval (S2, 1] is marked as severe; wherein S1 and S2 are classification thresholds, and 0<S1<S2≤1.

[0030] By counting the number of grayscale images marked as having reflection phenomena, the present invention can quantify the prevalence of the reflection problem, thereby more accurately understanding the overall quality status of the production line; by eliminating the normal overexposed areas in the normal detection image set, the real reflection problem areas can be more accurately identified to avoid misjudgment.

[0031] Preferably, the classification threshold includes a first threshold and a second threshold;

[0032] The method for obtaining the classification threshold includes:

[0033] Preset several key areas and determine whether the remaining reflective area includes the key areas;

[0034] If yes, then calculate the average percentage of the key areas in the remaining reflective areas of all grayscale images on the corresponding production line, and sort the average percentages in descending order, take the median of the arrangement as the second threshold S2, and take half of the median as the first threshold S1;

[0035] If not, the reflective condition of the corresponding bag will be marked as slight; the key areas include: seams, zippers, buckles and areas where decorative parts are located;

[0036] The calculation method for calculating the average proportion of the key areas contained in the remaining reflective area of ​​all grayscale images on the corresponding production line is: to obtain the proportion by dividing the number of key areas contained in the remaining area by the total number of key areas.

[0037] The present invention helps to avoid misjudging small-area reflections in non-critical areas as serious problems by identifying key areas, thereby improving the accuracy of assessment; using the median as the definition standard of the second threshold value S2 helps to reduce the impact of extreme values ​​on the assessment results; for the remaining reflective area that does not include the key area, its reflective condition is directly marked as slight, which helps to reduce misjudgment caused by over-interpretation of reflection problems.

[0038] Preferably, the determining whether the corresponding production line needs to add a visual sensor includes:

[0039] Obtain all grayscale images collected by the same visual sensor on the corresponding production line, and count the proportion of images marked as having reflections among all grayscale images; determine whether the image proportion is greater than a set quantity threshold; if yes, mark the corresponding visual sensor as a Class A sensor; if no, mark the corresponding visual sensor as a non-Class A sensor;

[0040] Count the number of all Class A sensors on the corresponding production line; determine whether the number of Class A sensors exceeds the preset sensor quantity threshold; if yes, mark the corresponding production line as not needing to add visual sensors; if no, mark the corresponding production line as needing to add visual sensors.

[0041] By counting the number of Class A sensors and comparing it with a preset sensor quantity threshold, the present invention can scientifically and reasonably evaluate the overall performance of sensors on the production line and provide a basis for increasing the number of sensors.

[0042] Preferably, the step of determining the number of visual sensors to be added to the corresponding production line according to the reflection situation includes:

[0043] Count the number of grayscale images under different reflective conditions; sum the number of grayscale images with slight, moderate and severe reflective conditions to obtain the total number;

[0044] The number of grayscale images under different reflective conditions is divided by the total number of images to obtain the proportion of reflective images under different reflective conditions; determine whether the proportion of images with slight reflective conditions exceeds 0.8;

[0045] If yes, the reflective grade D of the corresponding production line will be marked as Grade 1;

[0046] If no, it is determined whether the proportion of images with moderate reflectivity exceeds 0.5; if yes, the reflectivity level D of the corresponding production line is marked as level 2; if no, the reflectivity level D of the corresponding production line is marked as level 3;

[0047] Reflection compensation coefficient α through a preset grading decision table;

[0048] Obtain the sudden failure frequency f on the corresponding production line within the set period, the total cost S of all visual sensors on the corresponding production line, and the price sensitivity factor β; where the value range of the price sensitivity factor is [0.1, 0.5];

[0049] By formula The number of visual sensors added to the corresponding production line is calculated; where L Z is the length of the corresponding production line, L C is the coverage width of a single sensor, N o is the number of original visual sensors on the corresponding production line;

[0050] The sudden failure frequency is obtained by dividing the number of sudden failures within a set period by the total production batches to obtain the sudden failure frequency.

[0051] It should be noted that the price sensitivity factor is determined according to the value level of the corresponding luggage, and the value level is set by the staff; if the value level is high, the price sensitivity factor is marked as 0.5; if the value level is low, the price sensitivity factor is marked as 0.1; for other value levels, the corresponding price sensitivity factors are marked as 0.3.

[0052] The present invention can objectively and accurately evaluate the severity of the reflection problem on the production line by counting the number of grayscale images under different reflection conditions and calculating the image ratio of each reflection condition; setting the reflection level according to the ratio of reflective images helps to manage the reflection problem in a graded manner, thereby providing a basis for subsequently increasing the number of sensors.

[0053] Preferably, the step of determining whether the reflective condition after adding the visual sensor is degraded to a preset degree includes:

[0054] Retrieve the grayscale image after adding the visual sensor, and repeat step S200 to obtain a new reflection situation; when the new reflection situation is compared with the original reflection situation and the preset degree is degraded, the reflection situation after adding the visual sensor is marked as improved; otherwise, the reflection situation after adding the visual sensor is marked as not improved; wherein the preset degree of degradation is: the original reflection situation is severe or moderate, and the new reflection situation is mild; the original reflection situation is severe, and the new reflection situation is moderate; the original reflection situation is mild, and the new reflection situation is still mild.

[0055] To achieve the above-mentioned purpose, the second aspect of the present invention provides a monitoring and early warning system for bag production, comprising: a data acquisition module, a reflection judgment module and a sensor quantity determination module;

[0056] Data acquisition module: obtain normal detection image sets of different types of bags through pre-detection; collect grayscale images of different types of bags on the production line in real time through visual sensors;

[0057] Reflection judgment module: judge whether there is reflection in the grayscale image of the corresponding type of bags on the same production line;

[0058] Yes, the reflectivity of the luggage is determined based on the reflectivity in the grayscale image;

[0059] If not, determine whether the reflection of the corresponding type of bags on all production lines is obtained; if yes, mark it as end of determination; if not, re-determine whether there is reflection in the grayscale image of the corresponding type of bags on the same production line;

[0060] Sensor quantity determination module: determines whether the corresponding production line needs to add visual sensors;

[0061] If yes, determine the number of visual sensors to be added to the corresponding production line based on the reflection situation; determine whether the reflection situation after adding the visual sensors has degraded to a preset degree; if yes, stop adding the number of sensors and complete the production monitoring process; if no, send a reminder signal;

[0062] If not, a defect detection model is constructed to perform fault detection on the grayscale image with reflection phenomenon to obtain the fault type and issue a warning signal.

[0063] To achieve the above-mentioned purpose, the third aspect of the present invention provides a monitoring and early warning computing device for luggage production, comprising: a memory and a processor, wherein the memory stores executable instructions of the processor; wherein the processor is configured to execute a monitoring and early warning method for luggage production provided in the first aspect by executing the executable instructions.

[0064] Compared with the prior art, the present invention has the following beneficial effects:

[0065] 1. The present invention obtains a normal detection image set through pre-detection through the data acquisition module, which provides a benchmark for subsequent reflection judgment and defect detection, and helps to improve the accuracy of detection; by real-time acquisition of grayscale images, it can quickly respond to quality changes of bags on the production line and improve detection efficiency; through the reflection judgment module, it can automatically identify the reflection phenomenon in the grayscale image, reducing the dependence on manual detection, improving the automation level of detection, and determining the reflection situation of the bags according to the reflection phenomenon, which helps to further analyze the impact of reflection on the quality of bags.

[0066] 2. The present invention establishes a dynamic grayscale threshold through a normal image set under a standard viewing angle, eliminates misjudgments caused by material differences or process designs (such as metal decoration, etc.), reduces the false alarm rate of reflection detection, and dynamically calculates the number of visual sensors to be added according to the reflection situation to avoid over-deployment, thereby improving the reflection detection capability in a targeted manner and ensuring the stable operation of the production line; through reasonable configuration of visual sensors, the cost losses caused by production interruptions or quality problems caused by reflection problems can be reduced. At the same time, scientific decision-making methods can also help avoid excessive investment or waste of resources and reduce production costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0067] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0068] Figure 1 It is a schematic diagram of the monitoring and early warning process of the present invention;

[0069] Figure 2 This is a schematic diagram of the specific steps of image reflection judgment of the present invention;

[0070] Figure 3 A schematic diagram of the specific steps of determining the number of sensors and early warning according to the present invention;

[0071] Figure 4 This is a schematic diagram of the relationship between the modules included in the present invention. DETAILED DESCRIPTION

[0072] The technical scheme of the present invention will be clearly and completely described below in conjunction with the embodiments. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0073] See also Figure 1 The first embodiment of the present invention provides a monitoring and early warning method for bag production, comprising:

[0074] S100: obtaining normal inspection image sets of different types of bags through pre-inspection; collecting grayscale images of different types of bags on the production line in real time through visual sensors;

[0075] S200: Determine whether there is a reflection phenomenon in the grayscale image of the corresponding type of bags on the same production line;

[0076] Yes, the reflectivity of the luggage is determined based on the reflectivity in the grayscale image;

[0077] If not, determine whether the reflective conditions of the corresponding types of bags on all production lines are obtained; if yes, mark as end of determination; if not, repeat step S200;

[0078] S300: Determine whether the corresponding production line needs to add a visual sensor;

[0079] If yes, determine the number of visual sensors to be added to the corresponding production line based on the reflection situation; determine whether the reflection situation after adding the visual sensors has degraded to a preset degree; if yes, stop adding the number of sensors and complete the production monitoring process; if no, send a reminder signal;

[0080] If not, proceed to step S400;

[0081] S400: Construct a defect detection model to perform fault detection on the grayscale image with reflection phenomenon to obtain the fault type and issue a warning signal.

[0082] See also Figure 2, the specific steps of image reflection judgment are: obtaining the type of luggage to be detected and collecting luggage images under standard viewing angle; marking the luggage images with preset white areas of the luggage itself in the HSV color space; the standard viewing angle is: the viewing angle of the visual sensor collecting luggage images in the absence of reflective light sources in the production line;

[0083] Preprocess the bag image under the standard viewing angle to obtain a pre-grayscale image, and obtain the over-exposed grayscale value range corresponding to the white area of ​​the bag itself preset in the pre-grayscale image; use the pre-grayscale image within the over-exposed grayscale value range as a normal detection image set;

[0084] Retrieve a normal detection image set, and segment the normal overexposed area in the normal detection image set using the grayscale threshold segmentation method;

[0085] Retrieve grayscale images of different types of bags on the production line and count the overexposed areas within the overexposed grayscale range in the grayscale images;

[0086] Determine whether the overexposed area exceeds the corresponding normal overexposed area in the normal detection image set; if yes, mark the grayscale image of the corresponding type of luggage on the corresponding production line as having reflection phenomenon; if no, mark the grayscale image of the corresponding type of luggage on the corresponding production line as having no reflection phenomenon;

[0087] Count the number of all grayscale images marked as having reflections on the same production line; remove the normal overexposed area contained in the corresponding normal inspection image set from the overexposed area of ​​the grayscale image marked as having reflections to obtain the remaining reflection area; remove the normal overexposed area from the grayscale image to obtain the remaining area, and divide the remaining reflection area by the remaining area to obtain the remaining reflection area ratio;

[0088] Preset several key areas and determine whether the remaining reflective area includes the key areas;

[0089] If yes, then calculate the average percentage of the key areas in the remaining reflective areas of all grayscale images on the corresponding production line, and sort the average percentages in descending order, take the median of the arrangement as the second threshold S2, and take half of the median as the first threshold S1;

[0090] If not, the reflective condition of the corresponding bag will be marked as slight; the key areas include: seams, zippers, buckles and areas where decorative parts are located;

[0091] The reflective condition of the corresponding bags and luggage whose remaining reflective area ratio is in the interval (0, S1] is marked as slight;

[0092] The reflective condition of the corresponding bags and bags whose remaining reflective area ratio is in the interval (S1, S2] is marked as moderate;

[0093] The reflective situation of the corresponding bags and suitcases whose remaining reflective area ratio is in the interval (S2, 1] is marked as severe; wherein S1 and S2 are classification thresholds, and 0<S1<S2≤1.

[0094] For example, a bag factory needs to inspect bags on its production line, obtain the type of bags to be inspected, and collect bag images under a standard viewing angle; mark the bag images in the HSV color space to indicate that the hue H=0, the saturation S=0, and the brightness V are within the set interval [0.96, 1];

[0095] The bag image under the standard viewing angle is preprocessed to obtain a pre-grayscale image. By setting 0.96×255=244.8, 1×255=255, the over-exposed grayscale value range is the grayscale value in the interval [244.8, 255]. The pre-grayscale image in the over-exposed grayscale value range is taken as the normal detection image set.

[0096] Based on the adaptive threshold method, the normal over-exposure area in the normal detection image set is segmented by maximizing the inter-class variance and marked with color. The area of ​​the normal over-exposure area is 1260.21 square millimeters.

[0097] The grayscale images of the two types of bags A and B on the two production lines are collected in real time by visual sensors. One image is taken from each of the grayscale images of bags A and B and marked as T1 and T2. The resolution of the grayscale image is set to 1280×720 (the number of pixels is 921600). The grayscale image is calibrated such that 1 pixel = 0.1 mm. The number of pixels in the grayscale image T1 of type A with a grayscale value within [244.8, 255] is 126021, and the area is 126021×(0.1). 2 =1260.21 square millimeters; where T1 is the grayscale image with the highest grayscale value among the grayscale images of type A;

[0098] The number of pixels with grayscale values ​​in [244.8, 255] in the grayscale image T2 of category B is 214301, and the area is 314301×(0.1) 2 =3143.01 square millimeters;

[0099] Since 1260.21 square millimeters = 1260.21 square millimeters, the grayscale image T1 of the A-type luggage is marked as having no reflection phenomenon, and 3143.01 square millimeters is greater than 1260.21 square millimeters, the grayscale image T2 of the B-type luggage is marked as having reflection phenomenon;

[0100] There are 37 grayscale images marked as reflective on the production line of Class B bags. The remaining reflective area is 1882.8 square millimeters after subtracting 1260.21 square millimeters from 3143.01 square millimeters. The remaining area is 7955.79 square millimeters after subtracting 1260.21 square millimeters from 9216 square millimeters.

[0101] Divide the remaining reflective area by the remaining area to get the remaining reflective area ratio.

[0102] The number of key areas included in the preset Class B bags are: two seams, three zippers, one lock and two decorative parts;

[0103] The average percentage of key areas in the remaining reflective area is calculated by dividing the number of key areas in the remaining area by the total number of key areas, and the average percentages are arranged in descending order. The median of the arrangement is As the second threshold S2, half of the median, i.e. as a first threshold value S1;

[0104] The remaining reflective area accounts for 0.236 and is in the range If the reflection condition of the grayscale image T2 of the type B luggage is medium, the reflection condition of the grayscale image T2 of the type B luggage is marked as medium.

[0105] See also Figure 3 ,The specific steps of sensor quantity determination and early warning are: obtain all grayscale images collected by the same visual sensor on the corresponding production line, and count the proportion of images marked as having reflection phenomenon in all grayscale images; determine whether the image proportion is greater than the set quantity threshold; if yes, mark the corresponding visual sensor as a Class A sensor; if no, mark the corresponding visual sensor as a non-Class A sensor;

[0106] Count the number of all Class A sensors on the corresponding production line; determine whether the number of Class A sensors exceeds the preset sensor quantity threshold; if yes, mark the corresponding production line as not needing to add visual sensors; if no, mark the corresponding production line as needing to add visual sensors;

[0107] Count the number of grayscale images under different reflective conditions; sum the number of grayscale images with slight, moderate and severe reflective conditions to obtain the total number;

[0108] The number of grayscale images under different reflective conditions is divided by the total number of images to obtain the proportion of reflective images under different reflective conditions; determine whether the proportion of images with slight reflective conditions exceeds 0.8;

[0109] If yes, the reflective grade D of the corresponding production line will be marked as Grade 1;

[0110] If no, it is determined whether the proportion of images with moderate reflectivity exceeds 0.5; if yes, the reflectivity level D of the corresponding production line is marked as level 2; if no, the reflectivity level D of the corresponding production line is marked as level 3;

[0111] Reflection compensation coefficient α through a preset grading decision table;

[0112] Obtain the sudden failure frequency f on the corresponding production line within the set period, the total cost S of all visual sensors on the corresponding production line, and the price sensitivity factor β; where the value range of the price sensitivity factor is [0.1, 0.5];

[0113] By formula The number of visual sensors added to the corresponding production line is calculated; where L Z is the length of the corresponding production line, L C is the coverage width of a single sensor, N o is the number of original visual sensors on the corresponding production line;

[0114] Retrieve a grayscale image with reflection, and locate the reflective area through Fourier transform and high-pass filtering algorithm; detect the horizontal and vertical gradients in the reflective area through the Sobel operator, and obtain the contrast of the grayscale values ​​in the reflective area through the grayscale co-occurrence matrix; obtain the reflective feature expression through the contrast of the grayscale values, and input the reflective feature expression in the grayscale image into the defect detection model to obtain the fault type; wherein, the defect detection model is constructed according to the artificial intelligence model.

[0115] For example, a total of 100 grayscale images collected by the same visual sensor G1 on the corresponding production line are obtained. Among all the grayscale images, 15 images are marked as having reflections. The image proportion is There are 100 grayscale images collected by the visual sensor G2, and 10 images are marked as having reflections. The image ratio is There are 100 grayscale images collected by the visual sensor G3, and 12 images are marked as having reflections. The image ratio is

[0116] The preset quantity threshold is 0.1. Since 0.15 and 0.12 are both greater than 0.1, G1 and G2 are marked as Class A sensors;

[0117] The number of all Class A sensors on the Class B luggage production line is 3, and the preset sensor number threshold is 5; since 3 < 5, the Class B luggage production line is marked as needing to add visual sensors;

[0118] The total number of grayscale images with slight, moderate, and severe reflections is 37. Among them, the proportion of images with slight reflections is The percentage of images with moderate reflectivity is The proportion of images with severe reflection is because If it is greater than 0.5, the reflective level D of the Class B production line is marked as Level 2;

[0119] According to the preset grading decision table (such as Table 1), the reflection compensation coefficient α is 0.2;

[0120] Table 1 Classification decision table

[0121] Reflective grade D Reflection compensation coefficient α Level 1 0.1 Level 2 0.2 Level 3 0.3

[0122] Get the frequency of sudden failures on the Class B luggage production line in three months: The total cost S of all visual sensors on the corresponding production line is 5,000 yuan and the price sensitivity factor β is 0.3; the original number of visual sensors is 3; the length of the B-type luggage production line L Z The coverage width of a single sensor is L C 5 meters;

[0123] By formula The calculation shows that the number of visual sensors added to the corresponding production line is 7;

[0124] It should be noted that the shooting angles of the added visual sensors are inconsistent with those of the original visual sensors on the production line;

[0125] Retrieve the grayscale image with reflection phenomenon, locate the reflection area through Fourier transform and high-pass filtering algorithm; detect the horizontal and vertical gradients in the reflection area through Sobel operator, and obtain the contrast of grayscale value in the reflection area through grayscale co-occurrence matrix; obtain the reflection feature expression through the contrast of grayscale value, input the reflection feature expression in the grayscale image into the defect detection model to construct the reflection feature and fault correspondence table (as shown in Table 2) to obtain the fault type;

[0126] Table 2 Reflection characteristics and fault correspondence

[0127] Fault type Reflective characteristics Surface cracks Linear dark lines or brightness discontinuities within the reflective area Material deformation Reflective shapes are distorted and asymmetrically distributed Seam cracking Reflective covering seams and irregular shadows on edges Excessive wear The reflective area diffuses, accompanied by a decrease in the surrounding grayscale value

[0128] See also Figure 4 , the second aspect of the present invention provides a monitoring and early warning system for bag production, including: a data acquisition module, a reflection judgment module and a sensor quantity determination module;

[0129] Data acquisition module: obtain normal detection image sets of different types of bags through pre-detection; collect grayscale images of different types of bags on the production line in real time through visual sensors;

[0130] Reflection judgment module: judge whether there is reflection in the grayscale image of the corresponding type of bags on the same production line;

[0131] Yes, the reflectivity of the luggage is determined based on the reflectivity in the grayscale image;

[0132] If not, determine whether the reflection of the corresponding type of bags on all production lines is obtained; if yes, mark it as end of determination; if not, re-determine whether there is reflection in the grayscale image of the corresponding type of bags on the same production line;

[0133] Sensor quantity determination module: determines whether the corresponding production line needs to add visual sensors;

[0134] If yes, determine the number of visual sensors to be added to the corresponding production line based on the reflection situation; determine whether the reflection situation after adding the visual sensors has degraded to a preset degree; if yes, stop adding the number of sensors and complete the production monitoring process; if no, send a reminder signal;

[0135] If not, a defect detection model is constructed to perform fault detection on the grayscale image with reflection phenomenon to obtain the fault type and issue a warning signal.

[0136] An embodiment of the third aspect of the present invention provides a monitoring and early warning computing device for luggage production, comprising: a memory and a processor, wherein the memory stores executable instructions of the processor; wherein the processor is configured to execute a monitoring and early warning system for luggage production provided in the first aspect by executing the executable instructions.

[0137] Part of the data in the above formula is calculated by removing the dimension and taking its numerical value. The formula is a formula closest to the actual situation obtained by software simulation of a large amount of collected data; the preset parameters and preset thresholds in the formula are set by technical personnel in this field according to actual conditions or obtained through simulation of a large amount of data.

[0138] The working principle of the present invention is as follows: the present invention obtains a set of normal detection images of different types of luggage through pre-detection; collects grayscale images of different types of luggage on the production line in real time through visual sensors; judges whether there is a reflection phenomenon in the grayscale images of the corresponding type of luggage on the same production line; if yes, determines the reflection situation of the luggage according to the reflection phenomenon in the grayscale image; if no, judges whether the reflection situation of the corresponding type of luggage on all production lines is obtained; judges whether the corresponding production line needs to add visual sensors; determines the number of visual sensors to be added to the corresponding production line according to the reflection situation; judges whether the reflection situation after adding visual sensors is degraded to a preset degree; constructs a defect detection model to perform fault detection on the grayscale images with reflection phenomenon to obtain the fault type, and sends out an early warning signal.

[0139] The above embodiments are only used to illustrate the technical method of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical method of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical method of the present invention.

Claims

1. A monitoring and early warning method for bag production, characterized in that: include: S100: obtaining normal detection image sets of different types of bags through pre-detection; The visual sensor collects grayscale images of different types of bags on the production line in real time; S200: Determine whether there is a reflection phenomenon in the grayscale image of the corresponding type of bags on the same production line; Yes, the reflectivity of the luggage is determined based on the reflectivity in the grayscale image; If not, determine whether the reflective conditions of the corresponding types of bags on all production lines are obtained; if yes, mark as end of determination; if not, repeat step S200; S300: Determine whether the corresponding production line needs to add a visual sensor; If yes, determine the number of visual sensors to be added to the corresponding production line based on the reflection situation; determine whether the reflection situation after adding the visual sensors has degraded to a preset degree; if yes, stop adding the number of sensors and complete the production monitoring process; if no, send a reminder signal; If not, proceed to step S400; S400: Construct a defect detection model to perform fault detection on the grayscale image with reflection phenomenon to obtain the fault type and issue a warning signal.

2. A monitoring and early warning method for bag production according to claim 1, characterized in that: The method of obtaining a normal detection image set of different types of bags through pre-detection includes: Obtain the type of bags to be detected and collect bag images under standard viewing angles; mark the bag images with preset white areas in the HSV color space; the standard viewing angle is the viewing angle of the bag images collected by the visual sensor in the absence of reflective light sources in the production line; The luggage image under the standard viewing angle is preprocessed to obtain a pre-grayscale image, and the over-exposed grayscale value range corresponding to the preset white area of ​​the luggage itself in the pre-grayscale image is obtained; the pre-grayscale image within the over-exposed grayscale value range is used as the normal detection image set.

3. The monitoring and early warning method for bag production according to claim 1 is characterized in that: The determining whether there is a reflection phenomenon in the grayscale image of the corresponding type of bags on the same production line includes: Retrieve a normal detection image set, and segment the normal overexposed area in the normal detection image set using the grayscale threshold segmentation method; Retrieve grayscale images of different types of bags on the production line and count the overexposed areas within the overexposed grayscale range in the grayscale images; Determine whether the area of ​​the overexposed region exceeds the corresponding normal overexposed region in the normal detection image set; if yes, mark the grayscale image of the corresponding type of luggage on the corresponding production line as having reflection; if no, mark the grayscale image of the corresponding type of luggage on the corresponding production line as having no reflection.

4. A monitoring and early warning method for bag production according to claim 3, characterized in that: Determining the reflective condition of the luggage according to the reflective phenomenon in the grayscale image includes: Count the number of all grayscale images marked as having reflections on the same production line; remove the normal overexposed area contained in the corresponding normal inspection image set from the overexposed area of ​​the grayscale image marked as having reflections to obtain the remaining reflection area; remove the normal overexposed area from the grayscale image to obtain the remaining area, and divide the remaining reflection area by the remaining area to obtain the remaining reflection area ratio; A classification threshold is set, and the remaining reflective area ratio is classified according to the classification threshold to obtain the reflective condition of the luggage; the reflective condition of the luggage is: slight, moderate and severe.

5. A monitoring and early warning method for bag production according to claim 4, characterized in that: The classification threshold includes a first threshold and a second threshold; The method for obtaining the classification threshold includes: Preset several key areas and determine whether the remaining reflective area includes the key areas; If yes, then calculate the average percentage of the key areas in the remaining reflective area of ​​all grayscale images on the corresponding production line, and sort the average percentages in descending order, use the median in the arrangement as the second threshold, and use half of the median as the first threshold; If not, the reflective condition of the corresponding bag will be marked as slight; the key areas include: seams, zippers, buckles and areas where decorative parts are located; The calculation method for calculating the average proportion of the key areas contained in the remaining reflective area of ​​all grayscale images on the corresponding production line is: to obtain the proportion by dividing the number of key areas contained in the remaining area by the total number of key areas.

6. The monitoring and early warning method for bag production according to claim 1 is characterized in that: The determining whether the corresponding production line needs to add a visual sensor includes: Obtain all grayscale images collected by the same visual sensor on the corresponding production line, and count the proportion of images marked as having reflections among all grayscale images; determine whether the image proportion is greater than a set quantity threshold; if yes, mark the corresponding visual sensor as a Class A sensor; if no, mark the corresponding visual sensor as a non-Class A sensor; Count the number of all Class A sensors on the corresponding production line; determine whether the number of Class A sensors exceeds the preset sensor quantity threshold; if yes, mark the corresponding production line as not needing to add visual sensors; if no, mark the corresponding production line as needing to add visual sensors.

7. The monitoring and early warning method for bag production according to claim 1 is characterized in that: The method of determining the number of visual sensors to be added to the corresponding production line according to the reflection situation includes: Retrieve the reflection situation, quantify the reflection situation to obtain the reflection level D of the corresponding production line, and use the preset grading decision table to calculate the reflection compensation coefficient α; obtain the sudden failure frequency f on the corresponding production line within the set period, the total cost S of all visual sensors on the corresponding production line, and the price sensitivity factor β; where the value range of the price sensitivity factor is [0.1, 0.5]; By formula The number of visual sensors added to the corresponding production line is calculated; where L Z is the length of the corresponding production line, L C is the coverage width of a single sensor, N o is the number of original visual sensors on the corresponding production line; The sudden failure frequency is obtained by dividing the number of sudden failures within a set period by the total production batches to obtain the sudden failure frequency.

8. The monitoring and early warning method for bag production according to claim 1 is characterized in that: The determining whether the reflection condition after adding the visual sensor is degraded to a preset degree includes: Retrieve the grayscale image after adding the visual sensor, and repeat step S200 to obtain a new reflection situation; when the new reflection situation is compared with the original reflection situation and the preset degree is degraded, the reflection situation after adding the visual sensor is marked as improved; otherwise, the reflection situation after adding the visual sensor is marked as not improved; wherein the preset degree of degradation is: the original reflection situation is severe or moderate, and the new reflection situation is mild; the original reflection situation is severe, and the new reflection situation is moderate; the original reflection situation is mild, and the new reflection situation is still mild.

9. A monitoring and early warning system for luggage production, applied to a monitoring and early warning method for luggage production as claimed in any one of claims 1 to 8, characterized in that: include: Data acquisition module, reflection judgment module and sensor quantity determination module; Data acquisition module: obtain normal inspection image sets of different types of bags through pre-inspection; The visual sensor collects grayscale images of different types of bags on the production line in real time; Reflection judgment module: judge whether there is reflection in the grayscale image of the corresponding type of bags on the same production line; Yes, the reflectivity of the luggage is determined based on the reflectivity in the grayscale image; No, determine whether the reflective conditions of the corresponding types of bags on all production lines are obtained; If yes, it is marked as end of judgment; if no, it is re-judged whether there is reflection in the grayscale image of the corresponding type of bags on the same production line; Sensor quantity determination module: determines whether the corresponding production line needs to add visual sensors; If yes, determine the number of visual sensors to be added to the corresponding production line based on the reflection situation; determine whether the reflection situation after adding the visual sensors has degraded to a preset degree; if yes, stop adding the number of sensors and complete the production monitoring process; if no, send a reminder signal; If not, a defect detection model is constructed to perform fault detection on the grayscale image with reflection phenomenon to obtain the fault type and issue a warning signal.

10. A monitoring and early warning computing device for bag production, characterized in that: include: A memory and a processor, wherein the memory stores executable instructions of the processor; wherein the processor is configured to execute a monitoring and early warning method for bag production as described in any one of claims 1-8 by executing the executable instructions.

Images