Search

CN-122001986-A - Fiber component image acquisition method based on multi-view-field multi-focal-plane automatic scanning

CN122001986ACN 122001986 ACN122001986 ACN 122001986ACN-122001986-A

Abstract

The invention relates to the technical field of image processing, in particular to a fiber component image acquisition method based on multi-view-field multi-focal-plane automatic scanning, which comprises the following steps of pre-scanning a fiber scanning sample to obtain a preview image; the method comprises the steps of carrying out multi-view fiber space feature analysis according to a preview image so as to construct a self-adaptive acquisition strategy, carrying out multi-view multi-focal plane image acquisition on a fiber scanning sample based on the self-adaptive acquisition strategy to obtain multi-focal layer image data corresponding to each view field, reorganizing focal plane clear unit images according to the multi-focal layer image data, carrying out space alignment on focal plane clear unit images of adjacent view fields, and sequentially filling the aligned focal plane clear unit images of each view field into a preset panoramic canvas to generate a target fiber component image. According to the invention, by constructing a self-adaptive acquisition strategy matched with fiber distribution, fiber component image acquisition is realized, and the problems of low multi-view acquisition efficiency and morphological fracture at the splicing position are effectively solved.

Inventors

  • LI BING
  • LIU JUNWEI

Assignees

  • 深圳市菲雀兰博科技研究中心有限公司

Dates

Publication Date
20260508
Application Date
20260409

Claims (10)

  1. 1. The fiber component image acquisition method based on multi-view-field multi-focal-plane automatic scanning is characterized by comprising the following steps of: The method comprises the steps of S1, performing image pre-scanning on a fiber scanning sample to obtain a preview image, performing multi-view fiber space feature analysis according to the preview image to obtain fiber coverage rate and stacking height of each view field, and evaluating the stacking state of each view field based on the fiber coverage rate and the stacking height so as to construct a self-adaptive acquisition strategy; S2, performing multi-view-field multi-focal-plane image acquisition on the fiber scanning sample based on an adaptive acquisition strategy, and acquiring multi-focal-layer image data corresponding to each view field; Step S3, calculating definition characteristic values layer by layer for multi-focal layer image data of the same view field, screening a layer sequence number with the largest definition characteristic value at each pixel position in the multi-focal layer image data as a maximum response layer index; And S4, extracting overlapping areas in the focal plane clear unit images of the adjacent fields, performing spatial alignment on the focal plane clear unit images of the adjacent fields based on the overlapping areas, and sequentially filling the aligned focal plane clear unit images of the fields into a preset panoramic canvas to generate a target fiber component image.
  2. 2. The method for acquiring the fiber component image based on the multi-field multi-focal-plane automatic scanning of claim 1, wherein the step S1 comprises the steps of: s11, constructing a scanning space coordinate system according to a fiber scanning sample; Step S12, dividing a fiber scanning sample into grid view field units according to a preset coarse scanning grid spacing, and calculating X-axis coordinate values and Y-axis coordinate values of a central point of each grid view field unit in a scanning space coordinate system to form a pre-scanning view field central coordinate list; S13, performing image pre-scanning on a fiber scanning sample based on a pre-scanning view field center coordinate list, and collecting preview images of different focal planes of grid view field units in the Z-axis direction; And S14, performing field fiber space feature analysis on the preview image to obtain the reference focal plane height, stacking height and fiber coverage rate of the field corresponding to the grid field unit.
  3. 3. The multi-field multi-focal plane auto-scan based fiber component image acquisition method of claim 2, wherein performing field-of-view fiber space feature analysis on the preview image comprises: Calculating the sum of absolute values of gray differences of adjacent pixels in each preview image to obtain a definition evaluation value of each preview image; The definition evaluation values of all the preview images are arranged in a descending order, the definition evaluation value at the first position after the arrangement is selected as peak definition, and the preview image corresponding to the peak definition and the Z-axis position corresponding to the peak definition are recorded as the reference focal plane height of the view field; Calculating the product of the peak value definition and a preset scale factor to obtain a definition lower limit threshold; Counting the difference between the maximum value and the minimum value of the Z-axis position value corresponding to the preview image with the definition evaluation value larger than the definition lower limit threshold, and taking the difference as the stacking height of the view field; Performing binary segmentation on the preview image corresponding to the reference focal plane height to obtain a binary segmented image; And calculating the ratio of the number of fiber foreground pixels in the binary segmentation image to the total number of pixels in the preview image to obtain the fiber coverage rate of the field of view.
  4. 4. The method for acquiring the fiber component image based on the multi-field multi-focal-plane automatic scanning according to claim 2, wherein estimating the stacking state of each field of view based on the fiber coverage and the stacking height in step S1, thereby constructing the adaptive acquisition strategy comprises: Comparing the fiber coverage rate of the grid view field units with a preset coverage rate threshold, marking the grid view field units with the fiber coverage rate smaller than the preset coverage rate threshold as a single-layer area, marking the grid view field units with the fiber coverage rate larger than or equal to the preset coverage rate threshold as a stacking area, and generating an area type identifier; traversing the region type identifiers corresponding to the grid view field units, detecting whether the region type identifiers of the current grid view field unit and the adjacent grid view field units are consistent, and merging the grid view field units which are identical in region type identifier and are adjacent in space into a connected region; Performing plane fitting on the reference focal plane heights of all grid view field units in the communication areas to obtain inclined reference planes of all the communication areas, and taking the inclined reference planes as Z-axis scanning zero-position references of all view fields in the areas; Aiming at a communication area with an area type mark being a single-layer area, setting a basic layer scanning range and a sparse step length which take an inclined reference plane as a center, and generating single-layer acquisition parameters; Aiming at a communication area with the area type marked as a stacking area, setting an extended sweep range according to the maximum stacking height of each grid view field unit in the communication area, setting a dense step length, and generating stacking acquisition parameters.
  5. 5. The method for acquiring a fiber component image based on multi-field multi-focal plane auto-scan according to claim 4, wherein estimating the stacking state of each field of view based on the fiber coverage and the stacking height in step S1, thereby constructing an adaptive acquisition strategy further comprises: Acquiring a spatial distribution topological structure of a communication area; Planning a view field acquisition sequence based on a spatial distribution topological structure, determining a linking sequence among the areas according to the shortest distance between the connected areas, and generating a view field acquisition sequence; and carrying out data encapsulation on the single-layer acquisition parameters or the stacked acquisition parameters corresponding to the communication areas of the central coordinates of each grid view field unit in the view field acquisition sequence, and constructing a self-adaptive acquisition strategy containing a differential automatic scanning control instruction.
  6. 6. The method for acquiring a fiber component image based on multi-field multi-focal plane auto-scan according to claim 1, wherein step S3 comprises the steps of: S31, performing Laplace convolution operation on each layer of image of the same view field in multi-focal layer image data, calculating gradient amplitude values of each pixel position, and generating a gradient evaluation graph sequence of the view field; Step S32, traversing all gradient response values of the same pixel position in the gradient evaluation chart sequence, screening layer sequence numbers corresponding to the maximum gradient amplitude value, and generating a maximum response layer index; Step S33, based on the layer sequence number recorded at each pixel position in the maximum response layer index, indexing and extracting RGB pixel values at corresponding positions from the multi-focal layer image data, and recombining to generate a full-focal-plane texture map; And step S34, carrying out depth gray value normalization processing on the maximum response layer index, constructing a depth index image, and adding the depth index image serving as an independent data channel to the full-focal-plane texture image to synthesize a focal-plane clear unit image containing texture information and depth information.
  7. 7. The method for acquiring a fiber component image based on multi-field multi-focal plane automatic scanning of claim 6, further comprising, before synthesizing a focal plane sharp unit image containing texture information and depth information: Calculating a layer sequence number variance value of each pixel position in the maximum response layer index in a preset neighborhood window, marking the pixel position with the variance value larger than a preset variance threshold value as an outlier, and generating an outlier mask; counting the proportion of the number of outliers in the outlier mask to the total number of pixels to obtain the ratio of the outliers; Judging whether the occupation ratio of the outlier is larger than a preset correction trigger threshold, if so, sequentially performing median filtering and mode filtering on the maximum response layer index, eliminating isolated layer sequence jump points and filling a small-range layer sequence hole to generate a smooth layer index; Positioning the position of a pixel to be corrected in the smooth layer index based on the outlier mask, and rounding and filling the position of the pixel to be corrected by using a distance weighted average value of the sequence numbers of the effective pixel layers in the neighborhood to generate a correction layer index; Re-extracting RGB pixel values of a corresponding layer from multi-focal-plane image data according to the layer sequence number recorded in the correction layer index so as to correct the full-focal-plane texture map, updating the correction layer index to a data channel, and outputting an optimized focal-plane clear unit image; And if the outlier occupation ratio is smaller than or equal to the correction trigger threshold, not optimizing the focal plane clear unit image.
  8. 8. The method for acquiring the fiber component image based on the multi-field multi-focal-plane automatic scanning according to claim 6, wherein the step S4 of extracting the overlapping area in the focal-plane clear unit images of the adjacent fields of view, before spatially aligning the focal-plane clear unit images of the adjacent fields of view based on the overlapping area, comprises: respectively cutting out image blocks corresponding to the overlapping areas from focal plane clear unit images of adjacent view fields, respectively extracting characteristic points from the two image blocks, and performing similarity matching to generate an initial matching point pair set; Obtaining depth gray values corresponding to all matching points in an initial matching point pair set, calculating depth deviation values of two endpoints in the same matching point pair, removing the matching point pair with the depth deviation value larger than a preset depth continuity threshold value, and generating an effective matching point pair set; And calculating the translation amount and the rotation angle between adjacent fields of view based on the coordinate position relation of each matching point pair in the effective matching point pair set, and generating field of view registration parameters.
  9. 9. The method for collecting fiber component images based on multi-field multi-focal-plane automatic scanning according to claim 8, wherein before sequentially filling the aligned clear unit images of each field of view into a preset panoramic canvas in step S4, further comprising: mapping focal plane clear unit images of adjacent fields to a unified coordinate system according to field registration parameters, calculating absolute values of gray level differences of corresponding pixels in an overlapping area as texture difference values, and calculating absolute values of depth gray level differences of corresponding pixels as degree difference values; Carrying out cost value weighted summation processing on the texture difference value and the depth difference value to generate a fusion cost map; and searching a communication path which penetrates through the overlapped area and has the smallest accumulated pixel cost value in the fusion cost graph, and marking the communication path as an optimal suture line.
  10. 10. The method for collecting fiber component images based on multi-field multi-focal-plane automatic scanning according to claim 9, wherein sequentially filling the aligned clear unit images of each field of view into a preset panoramic canvas in step S4 comprises: expanding a preset eclosion width to two sides by taking the optimal suture line as a center to form an eclosion area, calculating the vertical distance between each pixel in the eclosion area and the optimal suture line, normalizing the vertical distance to be a mixed weight, Linearly weighting and mixing corresponding pixels in the focal plane clear unit images of adjacent fields according to the mixing weights to generate a smooth transition zone; And calculating absolute coordinate positions of all the fields in a preset panoramic canvas according to field registration parameters of all the fields, sequentially filling focal plane clear unit images of all the fields into the preset panoramic canvas according to the absolute coordinate positions, replacing original pixel values of an overlapping area with a smooth transition zone, and cutting out an incomplete area at the edge of the panoramic canvas to obtain a target fiber component image.

Description

Fiber component image acquisition method based on multi-view-field multi-focal-plane automatic scanning Technical Field The invention relates to the technical field of image processing, in particular to a fiber component image acquisition method based on multi-view-field multi-focal-plane automatic scanning. Background At present, the full-automatic microscopic imaging system is used for carrying out morphological recognition and quantitative analysis on textile fibers, so that manual microscopic examination is gradually replaced, and the method becomes a main technical means for detecting fiber components. Such systems typically consist of an motorized stage, a Z-axis focusing module, and a high resolution camera, which are controlled to move the stage field-by-field in the XY horizontal direction and to scan multiple layers in the Z-axis vertical direction to obtain microscopic image data of a fiber sample on a slide, for example, chinese patent CN223259959U. In the process of executing multi-view multi-focal-plane scanning, the prior art generally adopts a global fixed scanning strategy, namely, the same Z-axis layer number and range are preset for all view fields, however, the spatial distribution of fiber samples on a glass slide has obvious non-uniformity (such as dense central stacking and sparse and flat edges), the strategy lacking self-adaption capability causes that a large number of invalid redundant scanning is executed in sparse areas, and a complete focal plane is difficult to cover in dense stacking areas, so that invalid scanning time is overlong and overall acquisition efficiency is low, in addition, when a panoramic full-focal-plane composite image is synthesized, because the glass slide surface has small inclination and depth continuity constraint of crossing view fields is lacking, adjacent view fields are easy to select inconsistent focal planes to image in overlapped areas, and the deviation causes morphological fracture or local decoking of fibers crossing view fields at splicing positions, so that component characteristic identification distortion is caused, and the requirement of high-precision quantitative analysis is difficult to meet. Disclosure of Invention Based on the above, the present invention provides a fiber component image acquisition method based on multi-field multi-focal plane automatic scanning, so as to solve at least one of the above technical problems. In order to achieve the above purpose, a fiber component image acquisition method based on multi-view multi-focal plane automatic scanning comprises the following steps: The method comprises the steps of S1, performing image pre-scanning on a fiber scanning sample to obtain a preview image, performing multi-view fiber space feature analysis according to the preview image to obtain fiber coverage rate and stacking height of each view field, and evaluating the stacking state of each view field based on the fiber coverage rate and the stacking height so as to construct a self-adaptive acquisition strategy; S2, performing multi-view-field multi-focal-plane image acquisition on the fiber scanning sample based on an adaptive acquisition strategy, and acquiring multi-focal-layer image data corresponding to each view field; Step S3, calculating definition characteristic values layer by layer for multi-focal layer image data of the same view field, screening a layer sequence number with the largest definition characteristic value at each pixel position in the multi-focal layer image data as a maximum response layer index; And S4, extracting overlapping areas in the focal plane clear unit images of the adjacent fields, performing spatial alignment on the focal plane clear unit images of the adjacent fields based on the overlapping areas, and sequentially filling the aligned focal plane clear unit images of the fields into a preset panoramic canvas to generate a target fiber component image. Compared with the prior art, the method has at least the following advantages: According to the fiber component image acquisition method based on the multi-view multi-focal-plane automatic scanning, as the fiber scanning sample is subjected to image pre-scanning to obtain the preview image, the fiber coverage rate and the stacking height of each view field are obtained through multi-view fiber space feature analysis according to the preview image, and the stacking state of each view field is estimated based on the fiber coverage rate and the stacking height to construct the self-adaptive acquisition strategy. When the multi-view-field multi-focal-plane image acquisition is executed, firstly, the actual distribution characteristics of fibers on a slide glass are rapidly identified through pre-scanning, then Z-axis scanning parameters are configured differently according to the fiber coverage rate and stacking height of each view field, so that fewer scanning layers are adopted in a sparse area, sufficient scanning layers are adopted in a d