Image processing-based corrugated board automatic counting method and system

Abstract

The application relates to the technical field of image processing, in particular to a corrugated board automatic counting method and system based on image processing. The method comprises the following steps: acquiring a side image of a corrugated board stack, determining a boundary line set from the side image, and obtaining a sandwich space formed by adjacent boundary lines in the boundary line set; determining whether a pixel point on a dynamic central axis belongs to a solid state or an air state to obtain a binary state sequence for representing a transverse material distribution of the sandwich space; determining whether the sandwich space is a paperboard inner core layer or an inter-board isolation area according to the binary state sequence; and taking the sandwich space determined as the inter-board isolation area as a segmentation node to obtain counting result information of the corrugated board in the corrugated board stack. Through the above technical scheme, the number of layers of the corrugated board in the corrugated board stack can be automatically counted more accurately.

Description

Technical Field

[0001] This application relates to the field of image processing technology, and in particular to an automatic counting method and system for corrugated cardboard based on image processing. Background Technology

[0002] Corrugated cardboard is a multi-layer composite cardboard made of flat linerboard and corrugated core paper. As a core consumable in the modern logistics and packaging industry, corrugated cardboard is used for the transportation and protection of various commodities. In the production and delivery of corrugated cardboard, large quantities of corrugated cardboard are usually stacked and packaged into stacks for easy loading, unloading and transportation.

[0003] In related technologies, the number of stacked corrugated cardboard piles is usually counted manually or by simple photoelectric sensors. Photoelectric sensors use level transition signals generated by physical obstruction to count the cardboard that passes by continuously. The product can provide basic counting functions when handling cardboard of a single specification that moves continuously.

[0004] However, when faced with a tightly stacked corrugated cardboard, the physical boundaries of the side image are often blurred due to compression deformation or uneven lighting, making it difficult to accurately extract the material alternation features of the inner core layer, leading to misjudgment of the inner core layer and the isolation area between the boards.

[0005] Difficulties in extracting boundary features and chaotic internal structure analysis can lead to significant errors in the quantity counting of the entire stack of corrugated cardboard, resulting in quality incidents such as shortages or excesses in subsequent packaging. Incorrect counting results can disrupt the data accuracy of the inventory management system for warehousing and logistics, and increase the operating costs for enterprises in handling customer complaints and returns. Therefore, it is necessary to accurately achieve automatic counting of the number of layers of corrugated cardboard in a stack of corrugated cardboard. Summary of the Invention

[0006] To more accurately achieve automatic counting of the number of layers of corrugated cardboard in a stack of corrugated cardboard, this application provides an automatic counting method and system for corrugated cardboard based on image processing.

[0007] According to a first aspect of the embodiments of this application, an automatic counting method for corrugated cardboard based on image processing is provided, comprising: acquiring a side image of a stacked corrugated cardboard pile; determining a set of boundary lines from the side image to obtain a sandwich space formed by adjacent boundary lines in the boundary line set; vertically sorting the boundary lines in the boundary line set according to a height positioning reference; extracting the dynamic central axis of the sandwich space and determining a material determination threshold for the sandwich space; using the material determination threshold to determine whether the pixels on the dynamic central axis belong to a solid state or an air state, thereby obtaining a binarized state for characterizing the lateral material distribution of the sandwich space. The state sequence is used to obtain rhythmic feature values ​​to characterize the alternation of materials based on the distribution information of pixels belonging to the air state in the binarized state sequence. Using the rhythmic feature values ​​and the maximum hole width of the binarized state sequence, the interlayer space is determined to be the core layer inside the cardboard or the inter-board isolation area. Using the interlayer space determined to be the inter-board isolation area as the segmentation node, the side image of the corrugated cardboard stack is divided into multiple cardboard topological units. Using the number of core layers inside the cardboard in a single cardboard topological unit, the corrugated cardboard type of the single cardboard topological unit is determined to obtain the counting results of the corrugated cardboard in the corrugated cardboard stack.

[0008] In this way, compared to counting manually or by stacking height, the number of layers of corrugated cardboard in a stack can be counted more accurately and efficiently.

[0009] Optionally, determining the boundary line sequence from the side image includes: dividing the side image into multiple strips along the horizontal direction, extracting the average gray value of the pixels in each strip to obtain the vertical gray-scale distribution curve of the strip, and identifying the peak value in the vertical gray-scale distribution curve; when the width of the peak value is greater than a preset paper thickness threshold, determining a local concave point inside the peak value and splitting the peak value into two physical boundary positioning points; when the width of the peak value is less than or equal to the preset paper thickness threshold, determining the position corresponding to the peak value as a physical boundary positioning point; extracting the physical boundary positioning points within the strips, using a local linear regression algorithm to associate the physical boundary positioning points in adjacent strips to obtain continuous boundary lines, and sorting the continuous boundary lines according to the height positioning reference to obtain a boundary line sequence.

[0010] Optionally, the pixel on the dynamic central axis is determined to be either a solid state or an air state by using a material determination threshold, including: determining that the pixel belongs to a solid state when the gray value of the pixel on the dynamic central axis is greater than or equal to the material determination threshold; or determining that the pixel belongs to an air state when the gray value of the pixel on the dynamic central axis is less than the material determination threshold.

[0011] In this way, establishing a hard comparison logic based on grayscale values ​​and dynamic material determination thresholds helps to eliminate the interference of uneven global illumination on material state determination and enhances the system's adaptability to paperboards with different reflective properties.

[0012] Optionally, based on the distribution information of pixels belonging to the air state in the binarized state sequence, a rhythmic feature value for characterizing the material alternation pattern is obtained, including: determining segments composed of continuously distributed pixels belonging to the air state based on the distribution information of pixels belonging to the air state in the binarized state sequence to obtain the total number of segments; when the total number is less than or equal to a preset baseline number, the rhythmic feature value is determined as a preset penalty value; the preset penalty value is greater than or equal to the value used to determine the interlayer space as an inter-plate isolation area; when the total number is greater than the preset baseline number, the variance of the length of all segments in the binarized state sequence is used as the rhythmic feature value.

[0013] In this way, by introducing variance to measure the fluctuation of air segment length and using a preset penalty value for forced dimensionality reduction interception, we can suppress false regularity features caused by local damage and improve the objectivity of rhythm feature values ​​in characterizing corrugated wave structures.

[0014] Optionally, the interlayer space is determined to be the inner core layer of the cardboard or an inter-board isolation area by using the rhythm feature value and the maximum hole width of the binarized state sequence. This includes: determining the interlayer space as the inner core layer of the cardboard when the maximum hole width is less than a preset width, the rhythm feature value is less than a preset variance, and the vertical span of the interlayer space is greater than a preset lower limit of the core layer height; or, determining the interlayer space as an inter-board isolation area when the maximum hole width is greater than or equal to a preset width, or the rhythm feature value is greater than or equal to a preset variance, or the vertical span of the interlayer space is less than or equal to a preset lower limit of the core layer height.

[0015] Optionally, using the interlayer space identified as the inter-board isolation area as the dividing node, the side image of the corrugated cardboard stack is divided into multiple cardboard topological units, including: using the interlayer space identified as the inter-board isolation area as the dividing node, dividing the remaining interlayer space in the side image, excluding the inter-board isolation area, into multiple independent continuous spatial sequences; merging and combining the interlayer spaces belonging to the same continuous spatial sequence to divide the side image into multiple independent cardboard topological units.

[0016] This method of dividing the corrugated cardboard into sections based on isolation areas can accommodate the stacking of different types of corrugated cardboard, improving the counting accuracy when cardboard of different thicknesses is mixed and stacked.

[0017] Optionally, the corrugated board type of a single cardboard topology unit can be determined by the number of internal core layers in the cardboard within the single cardboard topology unit. This includes: determining the corrugated board type of a single cardboard topology unit by using the number of internal core layers in the cardboard within the single cardboard topology unit and a preset correspondence; the preset correspondence is used to characterize the relationship between different numbers of internal core layers and different corrugated board types.

[0018] Optionally, the counting results of corrugated cardboard are determined in the following way: the number of cardboard topological units of the same type of corrugated cardboard is summarized to obtain the number of corrugated cardboard of different types in the stacked corrugated cardboard pile, so as to output the counting results; the counting results are used to characterize the number of corrugated cardboard of different types.

[0019] In this way, mapping the number of core layers in the topological unit to the corrugated type and summarizing them by category helps to achieve accurate classification and inventory of mixed cardboard of multiple specifications, reducing the incomplete information defects caused by single quantity output.

[0020] Optionally, the method further includes: moving pixel by pixel along the boundary lines in the boundary line set in the horizontal direction to determine the difference in ordinate between the current pixel node and subsequent nodes separated by a preset horizontal step; locating extreme pixel points where the upward and downward trends on the boundary lines reverse during the pixel-by-pixel movement in the horizontal direction; extracting the leftmost starting coordinates and the rightmost ending coordinates of the boundary lines, and connecting the starting coordinates and the ending coordinates to generate a reference baseline; determining the shortest vertical spatial distance from the extreme pixel point to the reference baseline, and determining that the corresponding corrugated cardboard has warping deformation when the shortest vertical spatial distance is greater than a preset deviation.

[0021] Optionally, determining the material determination threshold of the interlayer space includes: taking the arithmetic mean of the gray values ​​of all pixels on the adjacent boundary lines constituting the interlayer space as the reflection reference value of the interlayer space, and taking the product of the reflection reference value and the preset internal light intensity attenuation coefficient as the material determination threshold of the interlayer space.

[0022] In this way, by dynamically generating a reflection reference value using the measured grayscale data of adjacent boundary lines and applying an attenuation coefficient adjustment, the effect of substrate darkening caused by insufficient light in the deep interlayer can be offset, thereby improving the matching degree between the threshold and the actual interlayer brightness.

[0023] Optionally, the dynamic central axis of the mezzanine space is extracted, including: traversing each pixel column in the mezzanine space, determining the first and second ordinates of the two adjacent boundary lines constituting the mezzanine space on the same pixel column; determining the midpoint ordinate of the vertical space of the pixel column based on the equal distribution of the sum of the first and second ordinates, and connecting the midpoint ordinates of all pixel columns to form the dynamic central axis.

[0024] Optionally, the boundary lines in the boundary line set are vertically sorted according to the height positioning reference, including: taking the arithmetic mean of the ordinates of pixels located on the same boundary line as the height positioning reference of the boundary line, and vertically sorting all boundary lines in ascending order of height positioning reference to obtain the boundary line set.

[0025] According to a second aspect of the present application, an automatic corrugated cardboard counting system based on image processing is provided, comprising: a processor and a memory, wherein the memory stores computer program instructions, and the computer program instructions, when executed by the processor, implement the steps of the automatic corrugated cardboard counting method based on image processing provided in the first aspect of the present application.

[0026] The technical solutions provided by the embodiments of this application may include the following beneficial effects: acquiring a side image of a stacked corrugated cardboard, being able to determine a set of boundary lines from the side image, and obtaining the interlayer space formed by adjacent boundary lines in the set of boundary lines; extracting the lateral material distribution features by combining the dynamic central axis of the interlayer space with the material determination threshold, so as to use the determined rhythm feature value and the maximum void width to distinguish the core layer inside the cardboard or the inter-board isolation area, thereby improving the accuracy of counting in complex stacking conditions.

[0027] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and do not limit this application. Attached Figure Description

[0028] Figure 1 This is a flowchart illustrating an automatic counting method for corrugated cardboard based on image processing, according to an exemplary embodiment. Figure 2 This is a schematic diagram showing the division of a single cardboard topological unit in a stack of corrugated cardboard; Figure 3 This is a schematic diagram illustrating the structure of an automatic corrugated cardboard counting system based on image processing, according to an exemplary embodiment. Detailed Implementation

[0029] To more accurately achieve automatic counting of the number of corrugated cardboard layers in a stack of corrugated cardboard, embodiments of this application provide an automatic corrugated cardboard counting method and system based on image processing. Figure 1 This is a flowchart illustrating an automatic counting method for corrugated cardboard based on image processing, according to an exemplary embodiment. Figure 1 As shown, the method includes the following steps.

[0030] In step S101, a side view of the stacked corrugated cardboard is obtained, and a set of boundary lines is determined from the side view to obtain the interlayer space formed by adjacent boundary lines in the set of boundary lines.

[0031] Industrial site image acquisition equipment can use industrial camera arrays to acquire side images of stacked corrugated cardboard. The camera's field of view can cover the entire material stacking area, and the side image contains panoramic layered texture information of the material from the bottom to the top.

[0032] The set of boundary lines can be determined from the side image. After extracting the set of boundary lines, the sandwich space formed by adjacent boundary lines in the set can be obtained based on the enclosing relationship between two adjacent boundary lines in the vertical space. The sandwich space reflects the physical gap between different paper layers. The boundary lines in the set of boundary lines are vertically sorted according to the height positioning reference.

[0033] In one embodiment, determining a boundary line sequence from a side image includes: dividing the side image into multiple strips along the horizontal direction; extracting the average gray value of pixels within each strip to obtain the vertical gray-scale distribution curve of the strip; identifying peaks in the vertical gray-scale distribution curve; when the width of a peak is greater than a preset paper thickness threshold, determining a local concave point inside the peak and splitting the peak into two physical boundary positioning points; when the width of a peak is less than or equal to the preset paper thickness threshold, determining the position corresponding to the peak as a physical boundary positioning point; extracting the physical boundary positioning points within the strips; using a local linear regression algorithm to associate the physical boundary positioning points in adjacent strips to obtain continuous boundary lines; and sorting the continuous boundary lines according to a height positioning reference to obtain a boundary line sequence.

[0034] The side image can be divided into multiple strips of fixed height along the horizontal direction, and the different strips are arranged along the vertical direction of the side image. The height of the strip can be set to, for example, 50 pixels. For each strip, the grayscale values ​​of the pixels can be read row by row along the vertical axis. By calculating the average grayscale value of all pixels in each row, a vertical grayscale distribution curve composed of different average grayscale values ​​can be obtained using the average grayscale values ​​corresponding to the pixels in different rows. The vertical grayscale distribution curve can characterize the changes in the average grayscale values ​​of pixels in different rows within the same strip.

[0035] The kraft paper face of corrugated cardboard has a high reflectivity under side lighting, which will form a significant high gray level protrusion on the vertical gray level distribution curve. The peak value in the vertical gray level distribution curve can be identified by the derivative zero-crossing point detection.

[0036] When two adjacent corrugated cardboard sheets are tightly bonded together, the two layers of face paper will merge into a wide bright area in the image. The horizontal pixel span of the peak in the vertical grayscale distribution curve can be detected. When the width of the peak is greater than the preset face paper thickness threshold, it can be determined that interlayer bonding has occurred. The preset face paper thickness threshold can be set to, for example, 8 pixels.

[0037] A slightly darker trough can be found in the center of the wide peak, and a local concave point can be determined inside the peak. The peak is then split into two physical boundary positioning points. For single-layer paper areas that have not been severely compressed, a narrow bright band will appear. When the width of the peak is less than or equal to the preset paper thickness threshold, there is no need to split the physical boundary positioning point. The vertex coordinates of the single peak are extracted as the representative position of the paper layer, and the position corresponding to the peak is determined as the physical boundary positioning point.

[0038] Isolated physical boundary points cannot represent the complete lateral extension shape of the cardboard. The physical boundary points within the strip can be extracted sequentially from left to right according to the horizontal direction of the strip. The physical boundary points in adjacent strips can be associated using a local linear regression algorithm to obtain continuous boundary lines. Local linear regression can smooth out coordinate jumps caused by burrs on the edge of the cardboard.

[0039] After obtaining multiple boundary lines, the vertical coordinate reference value of the line segments can be extracted, and the continuous boundary lines can be sorted according to the height positioning reference to obtain the boundary line sequence. Dividing the wide image into strips to extract peak values ​​and combining them with the width threshold for peak splitting can effectively deal with the boundary adhesion phenomenon caused by heavy pressure on corrugated cardboard and reduce the probability of missed detection.

[0040] In one embodiment, the boundary lines in the boundary line set are vertically sorted according to the height positioning reference, including: taking the arithmetic mean of the ordinates of pixels located on the same boundary line as the height positioning reference of the boundary line, and vertically sorting all boundary lines in ascending order of height positioning reference to obtain the boundary line set.

[0041] The boundary line is composed of multiple pixels spanning the width of the image. It can traverse all pixel nodes on a single boundary line and use the average value of the ordinate of pixels located on the same boundary line as the height positioning reference of the boundary line. This avoids the interference of coordinate extreme values ​​caused by local cardboard bending or edge curling, so that the height positioning reference can represent the overall height of the paper layer.

[0042] After obtaining the macroscopic height features of all boundary lines, the boundary lines can be vertically sorted according to the height positioning benchmark from smallest to largest to obtain a set of boundary lines. The height features are unified by using the arithmetic mean of the vertical coordinates and the sorting operation is performed from smallest to largest. This allows the creation of an index directory that is stacked from bottom to top, ensuring the continuity of the subsequent traversal process of the mezzanine space.

[0043] In step S102, the dynamic central axis of the interlayer space is extracted and the material determination threshold of the interlayer space is determined. The material determination threshold is used to determine whether the pixel points on the dynamic central axis belong to the solid state or the air state, so as to obtain a binary state sequence for characterizing the lateral material distribution of the interlayer space.

[0044] Because corrugated cardboard may be wavy and twisted after handling and getting damp, the interlayer space between the two boundary lines may not be horizontal. The dynamic central axis of the interlayer space can be determined by contour tracking extraction. The dynamic central axis can follow the bending trend of the cardboard and remain at the geometric center of the interlayer space. The material determination threshold of the interlayer space can be determined based on the local brightness characteristics of the image. The gray value of each pixel is extracted along the dynamic central axis, and the material determination threshold is used to determine whether the pixel on the dynamic central axis belongs to the solid state or the air state.

[0045] The corrugated core paper presents a continuous wavy structure on the side. The crests are filled with paper material and appear bright, while the troughs are filled with air and appear dark. The entire central axis can be binarized. By comparing the values, the grayscale information is converted into a sequence of 0s and 1s to obtain a binarized state sequence that represents the lateral material distribution of the interlayer space.

[0046] In one embodiment, extracting the dynamic central axis of the mezzanine space includes: traversing each pixel column within the mezzanine space, determining the first and second ordinates corresponding to the two adjacent boundary lines constituting the mezzanine space on the same pixel column; determining the midpoint ordinate of the vertical space of the pixel column based on the equal distribution of the sum of the first and second ordinates, and connecting the midpoint ordinates corresponding to all pixel columns to form the dynamic central axis.

[0047] The process can proceed horizontally in the two-dimensional coordinate system of the side image, traversing every pixel column within the interlayer space. At any pixel column, the upper and lower adjacent paper boundaries will intersect. By querying the coordinates, the two adjacent boundary lines constituting the interlayer space can be determined, along with the corresponding first and second ordinates on the same pixel column.

[0048] In order to find the center point of the interlayer and avoid the sampling trajectory from hitting the paper material at the upper and lower boundaries, the midpoint ordinate of the vertical space of the pixel column is determined based on the equal division result of the sum of the first and second ordinates. As the traversal process progresses to the rightmost side of the side image, the center position data of all columns can be obtained. The midpoint ordinates of all pixel columns are connected to form a dynamic central axis.

[0049] By calculating the midpoints column by column and connecting them using the summation and equal division of the vertical coordinates of the upper and lower boundaries, the sampling trajectory can be made to fit and adapt to the possible warping deformation of corrugated cardboard, thus avoiding the material misjudgment problem caused by sampling with a fixed horizontal line.

[0050] In one embodiment, determining the material determination threshold of the interlayer space includes: taking the arithmetic mean of the gray values ​​of all pixels on the adjacent boundary lines constituting the interlayer space as the reflection reference value of the interlayer space, and taking the product of the reflection reference value and a preset internal light intensity attenuation coefficient as the material determination threshold of the interlayer space.

[0051] Different height areas of the side image may be illuminated by external light sources at different angles, resulting in differences in overall brightness. By extracting the grayscale data of all pixels within the upper and lower boundary lines of the currently analyzed interlayer, the arithmetic mean of the grayscale values ​​of all pixels on the adjacent boundary lines constituting the interlayer space can be used as the reflection reference value of the interlayer space.

[0052] Because the interior of the interlayer has a deep, concave physical structure, light will be lost when it enters the interior, resulting in the actual reflective brightness of the inner corrugated paper being lower than that of the surface paper. A parameter matrix can be pre-stored in the memory, and the product of the reflection reference value and the preset internal light intensity attenuation coefficient can be used as the material determination threshold of the interlayer space.

[0053] For example, the preset internal light intensity attenuation coefficient can be set to 0.75. Specifically, it can be pre-calibrated according to the actual situation of the corrugated cardboard. The calibration process of the preset internal light intensity attenuation coefficient in this application embodiment will not be described in detail.

[0054] By dynamically generating a threshold using the gray-scale mean of the boundary line combined with the internal light intensity attenuation coefficient, adaptive adjustment can be achieved for different lighting areas and materials with different reflectivity, thus improving the accuracy of internal entity pixel extraction.

[0055] In one embodiment, determining whether a pixel on the dynamic central axis belongs to a solid state or an air state using a material determination threshold includes: determining that the pixel belongs to a solid state when the gray value of the pixel on the dynamic central axis is greater than or equal to the material determination threshold; or determining that the pixel belongs to an air state when the gray value of the pixel on the dynamic central axis is less than the material determination threshold.

[0056] The measured grayscale value of the pixel to be tested can be extracted along the dynamic central axis, and the measured grayscale value is compared with the material determination threshold obtained by the previous calculation. Since the physical material of corrugated core paper has a diffuse reflection effect on light, the local brightness is high. If the grayscale value of the pixel on the dynamic central axis is greater than or equal to the material determination threshold, it can be determined that the pixel belongs to the solid state.

[0057] The gaps inside the interlayer without cardboard filling absorb light, resulting in localized dark features. If the gray value of a pixel on the dynamic central axis is less than the material determination threshold, it can be determined that the pixel belongs to an air state.

[0058] In step S103, based on the distribution information of pixels belonging to the air state in the binarized state sequence, a rhythm feature value is obtained to characterize the material alternation pattern. Using the rhythm feature value and the maximum hole width of the binarized state sequence, the interlayer space is determined to be the core layer inside the cardboard or the inter-board isolation area.

[0059] The internal core of a normal corrugated cardboard is composed of continuous wavy base paper. The air gaps and solid core paper in the internal core of the corrugated cardboard will appear in a periodic alternation pattern. The entire sequence of data can be scanned, and the rhythmic feature value used to characterize the alternation pattern of materials can be obtained based on the distribution information of the pixels belonging to the air state in the binary state sequence.

[0060] In addition to the alternation pattern, the longest pixel distance that continuously appears as an empty state in the binarized state sequence can be extracted. By using the rhythm feature value and the maximum hole width of the binarized state sequence, the interlayer space can be determined to be the core layer inside the cardboard or the isolation area between boards.

[0061] When two corrugated cardboard sheets are stacked, the interlayer space between them is not made of corrugated paper. It not only lacks a periodic alternation pattern, but also often has a certain air gap. The difference in physical structure between the core layer of the cardboard or the interlayer space can be used to determine whether the interlayer space is the core layer of the cardboard or the interlayer space.

[0062] In one embodiment, obtaining a rhythmic feature value to characterize the alternation pattern of materials based on the distribution information of pixels belonging to the air state in the binarized state sequence includes: determining segments composed of continuously distributed pixels belonging to the air state based on the distribution information of pixels belonging to the air state in the binarized state sequence to obtain the total number of segments; when the total number is less than or equal to a preset baseline number, determining the rhythmic feature value as a preset penalty value; the preset penalty value is greater than or equal to the value used to determine the interlayer space as an inter-plate isolation area; when the total number is greater than the preset baseline number, using the variance of the lengths of all segments in the binarized state sequence as the rhythmic feature value.

[0063] A binarized state sequence is a one-dimensional digital array. Based on the distribution information of pixels belonging to the empty state in the binarized state sequence, we can find the intervals that are continuously empty state codes. Through connected component analysis algorithms, we can determine the segments composed of continuously distributed pixels belonging to the empty state, and obtain the total number of segments.

[0064] If the current interlayer is located in the gap between two pieces of cardboard, the current interlayer usually does not have a dense wave structure, resulting in very few air segments; if the total number is less than or equal to the preset baseline number, the rhythm characteristic value can be determined as the preset penalty value.

[0065] The preset penalty value is greater than or equal to the value used to determine the mezzanine space as an inter-board isolation area. For example, the preset penalty value can be set to a value that is larger than the distance between the inter-board isolation area, such as 99, to trigger the determination of the isolation area; the preset base number can be set to 10, for example.

[0066] For the actual internal wave core layer, there are more air segments. When the total number exceeds the preset baseline number, the number of pixels contained in each air segment can be extracted as the length feature. The periodic manufacturing process of the corrugated wave determines that the width of each hole in the corrugated wave is more similar. The mathematical average of the length of all air segments can be calculated, and then the dispersion of the length of each segment relative to the average value can be calculated.

[0067] The variance of the length of all segments in the binarized state sequence is used as the rhythm feature value. The smaller the variance, the more uniform the hole size, reflecting a standard corrugated structure. The number of air segments is used for judgment, and the length variance is introduced to measure the periodic stability. This can effectively shield the gaps in the cardboard that occasionally have a few holes due to damage, thus improving the robustness of structural feature extraction.

[0068] In one embodiment, determining the interlayer space as the inner core layer of the cardboard or an inter-board isolation area using rhythm feature values ​​and the maximum hole width of the binarized state sequence includes: determining the interlayer space as the inner core layer of the cardboard when the maximum hole width is less than a preset width, the rhythm feature value is less than a preset variance, and the vertical span of the interlayer space is greater than a preset lower limit of the core layer height; or, determining the interlayer space as an inter-board isolation area when the maximum hole width is greater than or equal to a preset width, or the rhythm feature value is greater than or equal to a preset variance, or the vertical span of the interlayer space is less than or equal to a preset lower limit of the core layer height.

[0069] Inter-board isolation area refers to the isolation area between different corrugated cardboards, while the internal core layer of cardboard refers to the internal structure of a single corrugated cardboard.

[0070] Since the size of a single hole in a real corrugated core is limited by the mold, the preset width can be set to accommodate the pixel span length of three standard corrugated crest cycles. For example, the preset width can be set to 150 pixels, and the preset variance can be determined in advance based on the variance of the pixels in the wave structure of a standard corrugated cardboard.

[0071] The vertical span of the interlayer space represents the vertical distance between the upper and lower physical boundaries, reflecting the height specification of the corrugated core layer inside a single corrugated cardboard. The preset core layer height can be predetermined based on the pixel distance of the corrugated core layer inside a single corrugated cardboard in the image. For example, the lower limit of the preset core layer height can be set to 4 pixels.

[0072] If the maximum void width is less than the preset width, the rhythm characteristic value is less than the preset variance, and the vertical span of the interlayer space is greater than the preset lower limit of the core layer height, the interlayer space can be determined to be the internal core layer of the cardboard.

[0073] If any dimension of the maximum void width, rhythm characteristic value, or vertical spatial span of the interlayer space does not conform to the characteristics of the corrugated wave structure, it indicates that a continuous air segment has appeared in the interlayer, which has disrupted the continuity constraint. Therefore, if the maximum void width is greater than or equal to the preset width, or the rhythm characteristic value is greater than or equal to the preset variance, it indicates that the hole distribution is extremely chaotic and does not possess the periodicity of mold processing.

[0074] When the vertical span of the mezzanine space is less than or equal to the lower limit of the preset core layer height, the mezzanine space can be determined as an inter-plate isolation area. Based on the multi-dimensional constraints of the maximum void width, rhythm characteristic variance, and vertical span, the accuracy of locating the inter-plate isolation area is improved.

[0075] In step S104, the interlayer space identified as the inter-board isolation area is used as the dividing node to divide the side image of the corrugated cardboard stack into multiple cardboard topology units. The corrugated cardboard type of a single cardboard topology unit is determined by the number of core layers inside the cardboard in the single cardboard topology unit, so as to obtain the counting result information of the corrugated cardboard in the corrugated cardboard stack.

[0076] After identifying all the layers, a catalog containing the inner core and interlayer isolation areas can be obtained. The area between two adjacent interlayer spaces identified as interlayer isolation areas represents a complete, independent corrugated cardboard entity. The area between the upper boundary of the corrugated cardboard stack and the interlayer space of the nearest interlayer isolation area, as well as the area between the lower boundary of the corrugated cardboard stack and the interlayer space of the nearest interlayer isolation area, both represent a complete layer of corrugated cardboard.

[0077] The interlayer space identified as the inter-board isolation area can be used as a dividing node to divide the side image of the corrugated cardboard stack into multiple cardboard topology units, each of which represents an actual material object.

[0078] For example, the current type of corrugated cardboard is three-layer corrugated cardboard, which includes boundary cardboard on both sides and a corrugated structure in the middle; the type of corrugated cardboard is five-layer corrugated cardboard, which includes boundary cardboard on both sides, two corrugated structures, and boundary cardboard between the corrugated structures.

[0079] By using the number of core layers inside a single cardboard topology unit, the corrugated cardboard type of that single cardboard topology unit can be determined. By traversing all topology units and summarizing the distribution data of corrugated cardboard types, the counting results of corrugated cardboard in a stack of corrugated cardboard can be obtained. For example, the counting results could be that the stack of corrugated cardboard includes three corrugated cardboards of the five-layer type.

[0080] In one embodiment, using the interlayer space identified as the inter-board isolation area as the dividing node, the side image of the corrugated cardboard stack is divided into multiple cardboard topological units, including: using the interlayer space identified as the inter-board isolation area as the dividing node, dividing the remaining interlayer space in the side image, excluding the inter-board isolation area, into multiple independent continuous spatial sequences; merging and combining the interlayer spaces belonging to the same continuous spatial sequence to divide the side image into multiple independent cardboard topological units.

[0081] The isolation area reflects the air gaps when the cardboard is stacked. Taking the interlayer space identified as the interlayer isolation area as the dividing node, the continuous context index of the interlayer catalog is cut off. The intermediate segment generated after being blocked by the isolation area is an overall structure composed of face paper and several internal core layers. The remaining interlayer space in the side image, except for the interlayer isolation area, can be divided into multiple independent continuous spatial sequences.

[0082] To restore the appearance of a single physical cardboard, a spatial merging operation can be performed to merge and combine the interlayer spaces belonging to the same continuous spatial sequence. The pixel regions after merging and combining constitute the analysis object containing the complete structure of the face paper and the corrugated paper, thereby dividing the side image into multiple independent cardboard topological units.

[0083] By utilizing the gap features of the inter-board isolation area to achieve the recombination of the internal interlayer, it can adapt to the detection scenario of mixed stacks of cardboard with different thicknesses and specifications, and avoid the problem of difficulty in obtaining relatively accurate segmentation results when using fixed pixel height and equal spacing for detection.

[0084] In one embodiment, determining the corrugated board type of a single cardboard topology unit by using the number of internal core layers in the cardboard in the single cardboard topology unit includes: determining the corrugated board type of the single cardboard topology unit by using the number of internal core layers in the cardboard in the single cardboard topology unit and a preset correspondence; the preset correspondence is used to characterize the relationship between different numbers of internal core layers and different corrugated board types.

[0085] A pre-stored mapping form can be invoked to determine the corrugated board type of a single cardboard topology unit by using the number of internal core layers in the cardboard and the preset correspondence. The mapping form records the classification criteria for corrugated board, and the preset correspondence is used to characterize the relationship between different numbers of internal core layers and different corrugated board types.

[0086] When the topological unit has only one core layer, it can be identified as a three-layer corrugated cardboard according to the preset correspondence. When the topological unit contains two core layers, it can be identified as a five-layer corrugated cardboard.

[0087] Based on the invocation of the statistical quantity of the core layer inside the topological unit and the preset correspondence, the automatic identification and classification of material specifications can be realized, so as to realize the detection of the type and quantity of corrugated cardboard in the stacked corrugated cardboard in the scenario of stacking multiple types of corrugated cardboard.

[0088] Figure 2 This is a schematic diagram showing the division of a single cardboard topological unit in a stack of corrugated cardboard, such as... Figure 2 As shown, there are multiple layers of corrugated cardboard stacked together, which effectively achieves the segmentation of each corrugated cardboard. It can avoid identifying the internal partitions in a single corrugated cardboard as partitions between different corrugated cardboards, and it can also avoid identifying the partitions between different corrugated cardboards as internal partitions of a single corrugated cardboard. It can achieve more accurate automated counting of corrugated cardboard, and can obtain more accurate and efficient counting results compared to manual counting or counting by stacking height.

[0089] In one embodiment, the counting result information of corrugated cardboard is determined by summarizing the number of cardboard topological units of the same type to obtain the number information of corrugated cardboard of different types in the stacked corrugated cardboard pile, so as to output the counting result information; the counting result information is used to characterize the number information of corrugated cardboard of different types.

[0090] After identifying the type of all individual corrugated cardboard topology units, the type labels of all topology units can be traversed, and the number of corrugated cardboard topology units with the same corrugated cardboard type can be summarized. For example, if it is found that there are 3 sheets of five-layer double-wall corrugated cardboard and 2 sheets of three-layer single-wall corrugated cardboard, a structured data report can be generated to obtain information on the quantity of different corrugated cardboard types in a stack of corrugated cardboard.

[0091] Reports containing categories and specific quantities can be pushed to the warehouse logistics management terminal to output counting results. The counting results are used to characterize the quantity of corrugated cardboard of different types. By classifying and summarizing, multi-dimensional counting statistics reports are output to meet the needs of intelligent warehousing for refined inventory and verification of materials.

[0092] In one embodiment, the system can also move horizontally pixel by pixel along the boundary lines in the boundary line set to determine the difference in ordinate between the current pixel node and subsequent nodes separated by a preset horizontal step; during the horizontal pixel-by-pixel movement, the system locates extreme pixel points where the upward and downward trends on the boundary lines reverse; the system extracts the leftmost starting coordinates and the rightmost ending coordinates of the boundary lines, and connects the starting coordinates and ending coordinates to generate a reference baseline; the system determines the shortest vertical spatial distance from the extreme pixel point to the reference baseline, and when the shortest vertical spatial distance is greater than a preset deviation, the system determines that the corresponding corrugated cardboard has warping deformation.

[0093] When corrugated cardboard is stored in a workshop with excessively high humidity, it is prone to structural warping and deformation, leading to packaging failure. In a two-dimensional pixel coordinate system, the data of each pixel coordinate point can be extracted by moving horizontally along the boundary line in the boundary line set pixel by pixel.

[0094] To capture the trend of the rising and falling slope of the line, the difference in the vertical coordinate between the current pixel node and the subsequent node separated by a preset horizontal step can be determined. The preset horizontal step can be set to, for example, 10 pixels.

[0095] If the corrugated cardboard deforms, the boundary line will show hill-like undulations and protrusions. At this time, the difference in the vertical coordinate will show a continuous positive cumulative increase. The continuous distance state of the positive increase can be monitored. During the process, when the difference in the vertical coordinate maintains a positive increase and the accumulated physical distance exceeds the preset span threshold, and then turns into a negative decrease and maintains the corresponding negative horizontal distance, the extreme pixel point where the rising and falling trend on the boundary line reverses is located. The extreme pixel point found is the highest protrusion position of the bending deformation. The preset span threshold can be set to, for example, 200 pixels.

[0096] To assess the severity of the protrusion, a flat reference benchmark can be established. The anchor points at both ends of the cardboard are obtained, and the coordinates of the leftmost starting point and the rightmost ending point of the boundary line are extracted. Using the principles of spatial geometry, the starting point coordinates and the ending point coordinates are connected in a two-dimensional spatial coordinate system to generate a reference benchmark line. The reference benchmark line represents the spatial distribution position that the cardboard should have in an absolutely flat state. The spatial deviation can be calculated using the formula for calculating the distance from a point to a line.

[0097] The shortest vertical spatial distance from the extreme pixel to the reference baseline can be determined. The shortest vertical spatial distance reflects the degree of warpage deviation. When the measured deviation exceeds the allowable error tolerance range of industrial packaging, that is, when the shortest vertical spatial distance is greater than the preset deviation, it can be determined that the corresponding corrugated cardboard has warpage deformation. The preset deviation can be set to 30 pixels.

[0098] By combining a virtual reference baseline with geometric distance comparison of reversal extreme points, non-contact automated quality inspection of cardboard flatness can be achieved, reducing machine jams and line stoppages caused by inferior products with severe warping and deformation flowing into subsequent automated packaging production lines.

[0099] Figure 3 This is a schematic diagram illustrating the structure of an image processing-based automatic counting system 1000 for corrugated cardboard, according to an exemplary embodiment. (Refer to...) Figure 3 The image processing-based automatic corrugated cardboard counting system 1000 includes a processor 1100 and a memory 1200. The memory 1200 stores computer program instructions, which, when executed by the processor 1100, implement all or part of the steps of the image processing-based automatic corrugated cardboard counting method of this application.

[0100] Other embodiments of this application will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of this application that follow the general principles of this application and include common knowledge or customary techniques in the art not disclosed herein. The specification and embodiments are to be considered exemplary only.

[0101] It should be understood that this application is not limited to the precise structure described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope.

Claims

1. An automatic counting method for corrugated cardboard based on image processing, characterized in that, include: Obtain a side view of the stacked corrugated cardboard, determine the set of boundary lines from the side view, and obtain the interlayer space formed by adjacent boundary lines in the set of boundary lines; the boundary lines in the set of boundary lines are vertically sorted according to the height positioning reference. Extract the dynamic central axis of the interlayer space and determine the material determination threshold of the interlayer space. Use the material determination threshold to determine whether the pixel on the dynamic central axis belongs to the solid state or the air state, so as to obtain a binary state sequence for characterizing the lateral material distribution of the interlayer space. Based on the distribution information of pixels belonging to the air state in the binarized state sequence, rhythmic feature values ​​are obtained to characterize the alternation pattern of materials. Using the rhythmic feature values ​​and the maximum hole width of the binarized state sequence, the interlayer space is determined to be the core layer inside the cardboard or the inter-board isolation area. Using the interlayer space identified as the inter-board isolation area as the dividing node, the side image of the corrugated cardboard stack is divided into multiple cardboard topological units. By using the number of core layers inside the cardboard in a single cardboard topological unit, the corrugated cardboard type of the single cardboard topological unit is determined, so as to obtain the counting results information of the corrugated cardboard in the corrugated cardboard stack.

2. The automatic counting method for corrugated cardboard based on image processing according to claim 1, characterized in that, Determining the boundary line sequence from the side view image includes: The side image is divided into multiple strips along the horizontal direction. The average gray value of the pixels in each strip is extracted to obtain the vertical gray value distribution curve of the strip. The peak value in the vertical gray value distribution curve is then identified. When the width of the peak is greater than the preset paper thickness threshold, a local concave point is determined inside the peak and the peak is split into two physical boundary positioning points; when the width of the peak is less than or equal to the preset paper thickness threshold, the position corresponding to the peak is determined as the physical boundary positioning point. Physical boundary positioning points within the strips are extracted. A local linear regression algorithm is used to associate the physical boundary positioning points in adjacent strips to obtain continuous boundary lines. The continuous boundary lines are then sorted according to the height positioning reference to obtain a boundary line sequence.

3. The automatic counting method for corrugated cardboard based on image processing according to claim 1, characterized in that, Determining whether pixels on the dynamic central axis belong to a solid state or an air state using a material determination threshold includes: If the grayscale value of a pixel on the dynamic central axis is greater than or equal to the material determination threshold, the pixel is determined to be in a solid state; or, if the grayscale value of a pixel on the dynamic central axis is less than the material determination threshold, the pixel is determined to be in an air state.

4. The automatic counting method for corrugated cardboard based on image processing according to claim 1, characterized in that, Based on the distribution information of pixels belonging to the air state in the binarized state sequence, rhythmic feature values ​​for characterizing the alternation pattern of materials are obtained, including: Based on the distribution information of pixels belonging to the air state in the binarized state sequence, determine the segments composed of continuously distributed pixels belonging to the air state to obtain the total number of segments. When the total number is less than or equal to the preset baseline number, the rhythm feature value is determined as the preset penalty value; the preset penalty value is greater than or equal to the value used to determine the interlayer space as the inter-plate isolation area; when the total number is greater than the preset baseline number, the variance of the length of all segments in the binarized state sequence is used as the rhythm feature value.

5. The automatic counting method for corrugated cardboard based on image processing according to claim 1, characterized in that, Using rhythmic feature values ​​and the maximum hole width of the binarized state sequence, the interlayer space is determined to be either the core layer inside the cardboard or an interlayer isolation area, including: If the maximum hole width is less than the preset width, the rhythm characteristic value is less than the preset variance, and the vertical space span of the interlayer space is greater than the preset core layer height lower limit, the interlayer space is determined to be the inner core layer of the cardboard. Alternatively, if the maximum void width is greater than or equal to the preset width, or the rhythm characteristic value is greater than or equal to the preset variance, or the vertical span of the mezzanine space is less than or equal to the preset lower limit of the core layer height, the mezzanine space is determined to be an inter-plate isolation area.

6. The automatic counting method for corrugated cardboard based on image processing according to claim 1, characterized in that, Using the interlayer space identified as the inter-board isolation area as the dividing node, the side view of the corrugated cardboard stack is divided into multiple cardboard topological units, including: Using the interlayer space identified as the inter-plate isolation area as the dividing node, the remaining interlayer space in the side image, excluding the inter-plate isolation area, is divided into multiple independent continuous spatial sequences. By merging and combining interlayer spaces belonging to the same continuous spatial sequence, the side image is divided into multiple independent cardboard topological units.

7. The automatic counting method for corrugated cardboard based on image processing according to claim 1, characterized in that, The corrugated board type of a single cardboard topology unit is determined by the number of core layers within the cardboard, including: The corrugated board type of a single cardboard topology unit is determined by the number of internal core layers in the cardboard and a preset correspondence. The preset correspondence is used to characterize the relationship between different numbers of internal core layers and different corrugated board types.

8. The automatic counting method for corrugated cardboard based on image processing according to claim 1, characterized in that, The counting results for corrugated cardboard are determined in the following ways: The number of topological units of the same type of corrugated cardboard is summarized to obtain the number of different types of corrugated cardboard in the stacked corrugated cardboard pile, so as to output the counting result information. The counting results information is used to characterize the quantity of corrugated board of different corrugated board types.

9. The automatic counting method for corrugated cardboard based on image processing according to claim 1, characterized in that, The method further includes: The system moves pixel by pixel along the boundary lines in the boundary line set in the horizontal direction to determine the difference in the vertical coordinate between the current pixel node and the subsequent node that is separated by a preset horizontal step; during the pixel-by-pixel movement in the horizontal direction, the system locates the extreme pixel points on the boundary lines where the upward and downward trends are reversed. Extract the starting coordinates of the leftmost point and the ending coordinates of the rightmost point of the boundary line, and connect the starting coordinates and the ending coordinates to generate a reference baseline; determine the shortest vertical spatial distance from the extreme pixel point to the reference baseline, and determine that the corresponding corrugated cardboard has warping deformation when the shortest vertical spatial distance is greater than the preset deviation.

10. An automatic counting system for corrugated cardboard based on image processing, characterized in that, include: A processor and a memory, the memory storing computer program instructions that, when executed by the processor, implement the image processing-based automatic counting method for corrugated cardboard according to any one of claims 1-9.

Images