Search

CN-121999194-A - Nursing pad printing toilet paper positioning method and system based on machine vision

CN121999194ACN 121999194 ACN121999194 ACN 121999194ACN-121999194-A

Abstract

The invention relates to the technical field of industrial machine vision, and discloses a method and a system for positioning nursing pad printing toilet paper based on machine vision. The method comprises the steps of fusing original vision, ambient light and equipment vibration information through multi-source sensing, and generating an anti-interference enhanced vision image. And on the enhanced image, carrying out region growth according to the gradient and texture constraint to obtain a candidate region. And extracting geometric and key spectral reflection characteristics of the region, matching with a standard template library, and calculating a space transformation parameter. And (3) finishing accurate spatial alignment of the candidate region and the template by applying the parameter, and finally realizing positioning by calculating the center of gravity and mapping coordinates. The method overcomes the illumination and vibration interference in the complex industrial environment, improves the distinguishing degree and accuracy of matching by utilizing the spectral characteristics, and realizes stable and reliable high-precision positioning.

Inventors

  • CHEN TAO
  • GAO XUHUA

Assignees

  • 山东爱舒乐卫生用品有限责任公司

Dates

Publication Date
20260508
Application Date
20251224

Claims (10)

  1. 1. A machine vision-based method for positioning a sanitary pad printing paper, the method comprising: Through a multisource sensing environment sensing process, synchronously capturing an original visual field information flow containing target nursing pad printing toilet paper, an ambient light interference field information flow and an equipment vibration disturbance information flow, and generating an enhanced visual characteristic image; performing seed point diffusion growth of a printing pattern area on the enhanced visual feature image, wherein the seed point diffusion growth performs area expansion according to the gray gradient and texture continuity constraint of the printing pattern until reaching a preset growth termination boundary, so as to outline a candidate positioning area; Extracting a geometric shape description subset and a spectral reflection description subset of the candidate positioning area, and matching the geometric shape description subset and the spectral reflection description subset with a standard nursing pad printing shape spectrum file pre-stored in a reference template library item by item; Calculating a space transformation parameter between the candidate positioning area and at least one standard nursing pad printing morphology spectrum file according to the matching result; carrying out rigidity adjustment on the coordinate frame of the candidate positioning area by using the space transformation parameters so as to enable the candidate positioning area to be spatially aligned with the selected standard nursing pad printing morphology spectrum file; And under the aligned coordinate frame, calculating the barycentric coordinates of the candidate positioning areas, and mapping the barycentric coordinates back to the originally captured visual field space to generate a final positioning decision for the nursing pad printing toilet paper.
  2. 2. The machine vision-based nursing pad printing toilet paper positioning method according to claim 1, wherein the synchronously capturing the original vision field information stream, the ambient light interference field information stream and the equipment vibration disturbance information stream containing the target nursing pad printing toilet paper through the multi-source sensing environment sensing process comprises: Performing space-frequency spectrum joint decomposition operation on the original visual field information flow, and stripping out a core visual characteristic wave band related to the printed pattern and a background visual characteristic wave band related to the toilet paper substrate; Fusing the ambient light interference field information flow and the equipment vibration disturbance information flow to generate a dynamic compensation vector field, and carrying out reverse disturbance calibration on the core visual characteristic wave band by utilizing the dynamic compensation vector field; Re-projecting and synthesizing the core visual characteristic wave band subjected to the reverse disturbance calibration and the background visual characteristic wave band to generate an enhanced visual characteristic image; The synchronization acquisition includes: a compound vision sensing head integrated by a wide-angle micro lens and an annular structured light projector is deployed, and the compound vision sensing head scans the printing toilet paper of the nursing pad on the production line at a fixed frequency; Starting the annular structured light projector, and projecting a grid-shaped light spot pattern with specific coding information on the surface of the printing toilet paper of the nursing pad at the moment of scanning; Synchronously receiving facula pattern images carrying surface deformation information and reflected by the surface of the nursing pad printing toilet paper through the wide-angle microscope lens, wherein the facula pattern images form the original visual field information flow; And installing an environment photometer and a triaxial micro-vibration sensor on the lateral side of the composite vision sensing head, and respectively and continuously recording illumination intensity change data and equipment vibration acceleration data in the scanning process to form an environment light interference field information flow and an equipment vibration disturbance information flow.
  3. 3. The machine vision-based nursing pad printing toilet paper positioning method as claimed in claim 2, wherein said performing a spatial-spectral joint decomposition operation on said original visual field information stream comprises: Converting continuous multi-frame images in the original visual field information flow into a frequency domain, and obtaining a spectrogram of each frame of image through Fourier transformation; Identifying in a spectrogram a periodic high frequency component caused by the grid-like spot pattern and a characteristic medium-low frequency component caused by the printed pattern and the substrate texture of the nursing pad printed toilet paper; designing a group of band-pass digital filters, separating and extracting the characteristic middle-low frequency components by using the band-pass digital filters, and reconstructing the characteristic middle-low frequency components into the core visual characteristic wave band after inverse Fourier transform; Subtracting the image data corresponding to the core visual characteristic wave band from the original visual field information stream, and removing the periodic high-frequency component noise from the rest part to form the background visual characteristic wave band.
  4. 4. The machine vision based nursing pad printing toilet paper positioning method as claimed in claim 2, wherein said generating a dynamic compensation vector field comprises: carrying out time sequence analysis on the ambient light interference field information flow, and extracting a low-frequency trend component of the ambient light interference field information flow which fluctuates along with time; Frequency analysis is carried out on the equipment vibration disturbance information flow, and main vibration frequency components and corresponding amplitudes of the equipment vibration disturbance information flow are identified; Establishing a two-dimensional mapping model taking time and image pixel positions as independent variables, taking the low-frequency trend component, the main vibration frequency component and the amplitude as input parameters, and calculating expected position offset and brightness fluctuation quantity of each pixel point at each time point caused by ambient light and vibration; The expected position offset and the amount of luminance fluctuation together comprise the motion compensation vector field, each vector in the motion compensation vector field comprising direction information and intensity information.
  5. 5. The machine vision-based nursing pad printing toilet paper positioning method as claimed in claim 4, wherein said performing inverse disturbance calibration on the core vision characteristic band using the dynamic compensation vector field comprises: reading time and position information corresponding to each pixel point of the core visual characteristic wave band in the dynamic compensation vector field; Performing inverse translation interpolation operation on the pixel point positions of the core visual characteristic wave bands in a spatial domain according to the direction information and the intensity information provided by the dynamic compensation vector field so as to offset the expected position offset; meanwhile, according to the brightness fluctuation amount provided by the dynamic compensation vector field, the pixel gray value of the core visual characteristic wave band is subjected to compensatory adjustment; And finishing the processing of all the pixel points, and outputting the core visual characteristic wave bands with corrected positions and brightness.
  6. 6. The machine vision-based nursing pad printing toilet paper positioning method as claimed in claim 1, wherein said performing seed point diffusion growth of a printed pattern area on the enhanced vision feature image comprises: on the enhanced visual feature image, automatically selecting at least one point as an initial seed point by adopting a local gray variance maximum point detection method; Calculating the consistency of gray gradient values and texture directions of pixel points in eight adjacent domains by taking the initial seed point as a center; setting a dynamic growth threshold, wherein the dynamic growth threshold is determined by the average gradient of edge pixels of the current growth area; The pixel points with gray gradient values smaller than the dynamic growth threshold and texture direction consistency higher than a preset consistency threshold in the eight neighborhood are brought into the current growth area, and the edges of the growth area are updated; Carrying out neighborhood exploration, threshold comparison and region updating processes iteratively until no new pixel is incorporated in a plurality of iterations or the edge of a growing region touches the physical boundary of the image, and judging that the growing termination boundary is reached at the moment; and marking all pixel areas covered by the final growth stopping areas as the candidate positioning areas.
  7. 7. The machine vision based nursing pad printing toilet paper positioning method as claimed in claim 6, wherein the extracting the geometrical morphology descriptor subset and the spectral reflectance descriptor subset of the candidate positioning area comprises: Performing contour tracking on the candidate positioning region to obtain a closed boundary curve of the candidate positioning region, calculating moment invariants of the boundary curve, fourier descriptors and curvature statistical characteristics, and combining the moment invariants, the Fourier descriptors and the curvature statistical characteristics into the geometrical morphology descriptor subset; And analyzing gray distribution of pixels in the candidate positioning area on the enhanced visual feature image under a plurality of narrow wave bands, and calculating the mean value, variance and high-order moment of the spectral reflectivity of the pixels to form the spectral reflection descriptor set.
  8. 8. The machine vision based nursing pad printing toilet paper positioning method as claimed in claim 1, wherein the item-by-item matching of the standard nursing pad printing topography spectrum files pre-stored in the reference template library comprises: Packaging the geometrical shape description subset and the spectrum reflection description subset into a characteristic vector; Calculating the mahalanobis distance between the feature vector and the standard feature vector corresponding to each standard nursing pad printing morphology spectrum file in the reference template library; Selecting a first plurality of standard nursing pad printing morphology spectrum files with minimum mahalanobis distance as matching candidates; and further calculating the local feature point corresponding relation between the candidate positioning area and each matching candidate by using an iterative nearest point algorithm, and screening out the standard nursing pad printing morphology spectrum file with optimal matching.
  9. 9. The machine vision based nursing pad printing toilet paper positioning method as set forth in claim 8, wherein calculating the spatial transformation parameters between the candidate positioning areas and at least one standard nursing pad printing topography spectrum profile based on the matching result includes: the spatial transformation parameters comprise rotation amount, translation amount and scale factor; Based on the local feature point corresponding relation between the candidate positioning area and the optimally matched standard nursing pad printing morphology spectrum file, solving an optimal rigid body transformation matrix by adopting a least square estimation method; -decomposing from said rigid body transformation matrix said rotation about an axis perpendicular to the image plane, said translation in the image plane and said scaling factor as a whole.
  10. 10. A machine vision-based nursing pad printing toilet paper positioning system comprising a memory, a processor and a computer program stored in the memory and running on the processor, characterized in that the processor, when executing the computer program, carries out the steps of the machine vision-based nursing pad printing toilet paper positioning method according to any one of the preceding claims 1 to 9.

Description

Nursing pad printing toilet paper positioning method and system based on machine vision Technical Field The invention relates to the technical field of industrial machine vision, in particular to a method and a system for positioning nursing pad printing toilet paper based on machine vision. Background On the automatic production line of sanitary products, the rapid and accurate visual positioning of the printed toilet paper of the nursing pad is a key link for realizing the follow-up automatic grabbing, cutting or packaging. The current mainstream approach is to capture a two-dimensional image of a single view using an industrial camera, followed by identification and localization of the printed area by edge detection, template matching, or color threshold based image segmentation techniques. The method can obtain a certain positioning effect under the ideal experimental environment of controlled stable illumination and no remarkable equipment vibration. In complex industrial scenarios where the actual high speed operation is performed, there is an uneven, flickering or random variation of the ambient light source, and the continuous operation of the production equipment may also introduce non-negligible high or low frequency mechanical vibrations. These factors act together on the image sensor, which causes problems such as blurring, noise or local overexposure/underexposure in the captured original image, and seriously interferes with the stability and accuracy of subsequent feature extraction. In addition, for different products with similar textures and colors or the situation that the contrast between the printed pattern and the background is not high, the printed pattern is matched only by relying on the traditional geometric shape or color characteristics, so that mismatching or positioning drift is very easy to generate, and the flexibility and the reliability of the production line are restricted. There is a need for a positioning technique that overcomes the dynamic interference of the environment and utilizes richer features for accurate identification. Disclosure of Invention The invention aims to provide a machine vision-based positioning method and system for a nursing pad printing toilet paper, which are used for solving the problems in the background technology. In order to achieve the above object, the present invention provides a machine vision-based positioning method for a sanitary pad printing paper, the method comprising: Through a multisource sensing environment sensing process, synchronously capturing an original visual field information flow containing target nursing pad printing toilet paper, an ambient light interference field information flow and an equipment vibration disturbance information flow, and generating an enhanced visual characteristic image; performing seed point diffusion growth of a printing pattern area on the enhanced visual feature image, wherein the seed point diffusion growth performs area expansion according to the gray gradient and texture continuity constraint of the printing pattern until reaching a preset growth termination boundary, so as to outline a candidate positioning area; Extracting a geometric shape description subset and a spectral reflection description subset of the candidate positioning area, and matching the geometric shape description subset and the spectral reflection description subset with a standard nursing pad printing shape spectrum file pre-stored in a reference template library item by item; Calculating a space transformation parameter between the candidate positioning area and at least one standard nursing pad printing morphology spectrum file according to the matching result; carrying out rigidity adjustment on the coordinate frame of the candidate positioning area by using the space transformation parameters so as to enable the candidate positioning area to be spatially aligned with the selected standard nursing pad printing morphology spectrum file; And under the aligned coordinate frame, calculating the barycentric coordinates of the candidate positioning areas, and mapping the barycentric coordinates back to the originally captured visual field space to generate a final positioning decision for the nursing pad printing toilet paper. Preferably, the step of synchronously capturing the original visual field information flow, the ambient light interference field information flow and the equipment vibration disturbance information flow containing the target nursing pad printing toilet paper through the multi-source sensing environment sensing process comprises the following steps of: Performing space-frequency spectrum joint decomposition operation on the original visual field information flow, and stripping out a core visual characteristic wave band related to the printed pattern and a background visual characteristic wave band related to the toilet paper substrate; Fusing the ambient light interference field information flow and