Search

CN-121685985-B - Glaze texture structured feature extraction method based on polarization difference and band fusion

CN121685985BCN 121685985 BCN121685985 BCN 121685985BCN-121685985-B

Abstract

The invention discloses a method for extracting structured features of a glaze pattern based on polarization difference and band fusion, and belongs to the technical field of image processing. The method comprises the steps of collecting a plurality of images of the same pattern under different polarization angles and different illumination wave bands, firstly registering the multi-frame images, then calculating polarization parameters and detecting a highlight region, then estimating and inhibiting specular reflection components based on a polarization degree physical model to obtain a highlight pattern removing image, then carrying out color and scale normalization processing on the image, finally segmenting and refining a pattern skeleton from the normalized image, extracting colors, curves, periods and texture features, and fusing to generate a structured feature code. The invention can effectively inhibit highlight interference, improve the definition and readability of the pattern details, realize the consistency and reproducibility of data under different acquisition conditions, output retrievable structural features and greatly improve the efficiency of digitally recording, managing and researching the glaze pattern.

Inventors

  • Lei Wenxin
  • GAO JIYANG
  • Ming Xiaochuan

Assignees

  • 齐鲁工业大学(山东省科学院)

Dates

Publication Date
20260508
Application Date
20260206

Claims (10)

  1. 1. The method for extracting the structured characteristics of the glaze texture based on the fusion of the polarization difference and the wave band is characterized by comprising the following steps: S1, acquiring an original image set of the same glaze pattern area under a plurality of groups of imaging conditions, wherein the plurality of groups of imaging conditions at least comprise sequences with different polarization angles and sequences with different illumination wave bands; S2, carrying out image registration on the original image set so that the same physical point in all images is mapped to the same pixel coordinate to obtain a registered image set; S3, calculating polarization parameters of each pixel based on images with different polarization angles in the registered image set; S4, generating a binary mask for marking a specular highlight area according to the polarization parameters and the pixel intensity; s5, estimating and removing specular reflection components from the registered image set based on the polarization parameters and the binary mask to obtain a deghosted pattern image; S6, carrying out color channel correction on the highlight removing pattern image, and outputting a standard pattern image with normalized color and scale; S7, carrying out pattern region segmentation and skeletonization on the standard pattern image, and extracting a geometric skeleton of the pattern; and S8, extracting and fusing color, curve, period and texture features based on the standard pattern image and the geometric skeleton, and generating the structural features of the glaze pattern.
  2. 2. The method for extracting structured features of a glaze pattern based on fusion of polarization differences and wavelength bands according to claim 1, wherein in step S1, the sequence of different polarization angles includes images with polarization directions of 0 °, 45 °, 90 ° and 135 °, and the sequence of different illumination wavelength bands includes images acquired under illumination of light sources of at least two different wavelengths.
  3. 3. The method for extracting the structured feature of the glaze texture based on the fusion of the polarization difference and the wave band according to claim 1, wherein the step S2 specifically comprises: S21, selecting a frame of image from the original image set as a reference frame; S22, extracting characteristic points between the rest of each frame of images and the reference frame and matching to obtain a matching point pair set; s23, solving affine transformation model parameters through a robust estimation algorithm based on the matching point pair set; s24, transforming the rest frame images under the coordinate system of the reference frame by utilizing the affine transformation model parameters, and finishing registration.
  4. 4. The method for extracting structured feature of glaze pattern based on fusion of polarization difference and band according to claim 1, wherein in step S3, the polarization parameters include stokes parameters 、 The calculation formula of the Stokes parameter is as follows: ; ; ; Wherein, the , , , The intensity values at pixel x of images having polarization angles of 0 °, 45 °, 90 °, 135 ° after registration are respectively represented.
  5. 5. The method for extracting the structured feature of the glaze texture based on the fusion of the polarization difference and the wave band according to claim 4, wherein the linear polarization degree DoLP and the polarization angle AoP of each pixel are calculated according to the stokes parameter, and the calculation formula is as follows: ; ; Wherein the method comprises the steps of Is a four-quadrant arc tangent, Is the degree of polarization of the pixel, Is the principal direction of polarization of the pixel, Is a very small positive number.
  6. 6. The method for extracting the structured feature of the glaze texture based on the fusion of the polarization difference and the wave band according to claim 4, wherein the step S4 specifically comprises: calculating a local luminance adaptive threshold value of the pixel x: ; Wherein, the Respectively by The average value and standard deviation of brightness in a local window are taken as the center, and alpha is a preset experience coefficient; defining highlight candidate masks The method comprises the following steps: ; ; wherein L (x) is the intensity of pixel x, Is a preset threshold value of polarization degree.
  7. 7. The method for extracting the structured feature of the glaze texture based on the fusion of the polarization difference and the wave band according to claim 1, wherein the step S5 specifically comprises: s51, selecting an image with minimum intensity of each pixel point from the registered image set As a base image; s52, estimating specular reflection components according to the linear polarization degree DoLP (x) and the intensity L (x): ; Wherein beta is a preset weight coefficient; s53, in the area marked by the binary mask as highlight, forming an image from the substrate Subtracting the specular component from And cut off the result to obtain a highlight pattern removed image : ; Wherein, the Is a truncated function.
  8. 8. The method for extracting the structured feature of the glaze texture based on the fusion of the polarization difference and the wave band according to claim 1, wherein the step S6 specifically comprises: s61, aiming at the highlight pattern removing image C e { R, G, B }, calculates its average intensity outside the highlight region ; S62, calculating gain coefficients of all channels: ; Wherein, the As the target gray-scale average value, Is an extremely small positive number; s63, pixel value of each channel Multiplying the corresponding gain coefficients to obtain a color normalized image : 。
  9. 9. The method for extracting the structured feature of the glaze texture based on the fusion of the polarization difference and the wave band according to claim 1, wherein the step S7 specifically comprises: S71, calculating the gradient amplitude of the standard pattern image, and obtaining a binary mask of the pattern region through threshold segmentation ; S72, performing mask alignment on the binary mask And (3) applying a thinning algorithm, and iteratively deleting boundary pixels meeting a preset deleting condition until a single-pixel wide pattern skeleton image Sk is obtained.
  10. 10. The method for extracting structured features of a glaze pattern based on polarization difference and band fusion according to claim 1, wherein in step S8, the structured features F are formed by splicing the following sub-feature vectors: Color characterization Which is a color histogram of the standard grain image; Curve characteristics The histogram is a histogram of the direction angles of all pixel points on the geometric framework; Periodic characteristics The method comprises the steps of calculating an autocorrelation function of a one-dimensional projection signal along the edge trim direction of the pattern, and extracting a period length L corresponding to the main peak position of the autocorrelation function; Texture features Which is a local binary pattern histogram of the standard pattern image.

Description

Glaze texture structured feature extraction method based on polarization difference and band fusion Technical Field The invention relates to the technical field of image processing, in particular to a method for extracting structured characteristics of a glaze pattern based on polarization difference and band fusion. Background The existing digital acquisition and analysis of the glazed ceramic pattern usually depends on the common camera/mobile phone shooting to be matched with external light supplementing, polarizing plates or light sheds and other equipment, and is processed through manual picture repair or a general highlight removing algorithm after acquisition. However, due to strong specular reflection on the glaze, the high-light position quickly drifts along with the change of the illumination direction and the shooting angle, so that the fine lines, gold drawing and dark lines of the pattern are often shielded, and meanwhile, the color temperature difference, the exposure strategy and the change of the shooting distance of light sources in different places can cause the non-standardization of the color and the scale, so that the result of the same pattern under different acquisition conditions is inconsistent, and the pattern is difficult to reproduce and transversely compare. The existing anti-reflection/anti-highlight method mainly comprises single frame image threshold restoration or simple fusion, lacks a mechanism for reliably estimating and inhibiting specular reflection components based on imaging condition differences such as polarization angle differences, multiband differences and the like, is easy to cause false smearing, boundary distortion or residual highlight of pattern details, and is easy to output enhanced pictures, and the existing process is lack of structural analysis and retrievable coding of pattern elements such as a decoration period, a curve skeleton, a hierarchical structure and the like, so that the technical requirements of 'reproducible, comparable and retrievable' of cultural relic record, pattern research and digital management are difficult to meet. Therefore, how to provide a method for extracting structured features of a glazed pattern, which can improve the consistency, reliability and application efficiency of pattern data, is a technical problem that needs to be solved by those skilled in the art. Disclosure of Invention The present invention has been made in view of the above problems, and has as its object to provide a method for extracting structured features of a glaze pattern based on fusion of polarization differences with bands, which overcomes or at least partially solves the above problems. In order to achieve the above purpose, the present invention adopts the following technical scheme: The embodiment of the invention provides a method for extracting structured characteristics of a glaze pattern based on fusion of polarization difference and wave bands, which comprises the following steps: S1, acquiring an original image set of the same glaze pattern area under a plurality of groups of imaging conditions, wherein the plurality of groups of imaging conditions at least comprise sequences with different polarization angles and sequences with different illumination wave bands; S2, carrying out image registration on the original image set so that the same physical point in all images is mapped to the same pixel coordinate to obtain a registered image set; S3, calculating polarization parameters of each pixel based on images with different polarization angles in the registered image set; S4, generating a binary mask for marking a specular highlight area according to the polarization parameters and the pixel intensity; s5, estimating and removing specular reflection components from the registered image set based on the polarization parameters and the binary mask to obtain a deghosted pattern image; S6, carrying out color channel correction on the highlight removing pattern image, and outputting a standard pattern image with normalized color and scale; S7, carrying out pattern region segmentation and skeletonization on the standard pattern image, and extracting a geometric skeleton of the pattern; and S8, extracting and fusing color, curve, period and texture features based on the standard pattern image and the geometric skeleton, and generating the structural features of the glaze pattern. Preferably, in step S1, the sequence of different polarization angles comprises images with polarization directions of 0 °, 45 °,90 ° and 135 °, and the sequence of different illumination bands comprises images acquired under illumination of at least two light sources of different wavelengths. Preferably, step S2 specifically includes: S21, selecting a frame of image from the original image set as a reference frame; S22, extracting characteristic points between the rest of each frame of images and the reference frame and matching to obtain a matching point pair set; s23,