CN-122023290-A - Biomass particle finished product quality grading method and system based on image processing

CN122023290ACN 122023290 ACN122023290 ACN 122023290ACN-122023290-A

Abstract

The application provides a quality grading method and a quality grading system for biomass particle finished products based on image processing, and relates to the technical field of quality grading of biomass particle finished products. Then matching the identification result with a preset damage characteristic database to obtain an initial damage rate, and dynamically correcting the initial value by combining the particle number statistics and the size distribution characteristics of the sampling sample; finally, the corrected breakage rate is compared with a preset quality interval, so that the automatic quality grade division of the biomass particle finished product is realized. The method effectively solves the technical problems that the damage of the three-dimensional surface and the representativeness of a sample are difficult to accurately evaluate in the traditional two-dimensional detection through the cooperative application of the three-dimensional vision technology and the statistical correction model.

Inventors

YANG HUI
MAO WEI
MAO XIAOHONG

Assignees

江山华隆能源开发有限公司

Dates

Publication Date: 20260512
Application Date: 20260114

Claims (10)

1. The biomass particle finished product quality grading method based on image processing is characterized by comprising the following steps of: Combining left eye images and right eye images of sampling samples of preset biomass particles at different angles to obtain binocular vision image data; parallax calculation is carried out on the binocular vision image data, and three-dimensional contour data of the preset biomass particles are generated; performing image recognition processing on the three-dimensional profile data to recognize a damaged area of the preset biomass particles; matching preset damage characteristics with damage rate associated data and the damage area to obtain a surface damage rate initial value; the initial value of the surface breakage rate is adjusted by using the particle number of the sampling sample and the particle size distribution of the sampling sample, so that the adjusted surface breakage rate is obtained; and comparing the adjusted surface breakage rate with a preset breakage rate interval to divide the quality grade of the finished product of the preset biomass particles according to the comparison result.
2. The method of claim 1, wherein said adjusting the initial value of the surface breakage rate using the number of particles of the sampled sample and the particle size distribution of the sampled sample to obtain the adjusted surface breakage rate comprises: calculating a difference value between the particle number of the sampling sample and a preset reference number as a first difference value, and taking an absolute value of the first difference value as a number difference value; Counting the share of the number of the large-size particles in the sampling sample to the number of the particles according to the particle size distribution of the sampling sample to be used as the share of the large-size particles; calculating a difference value between the large-size particle share and a preset share reference as a second difference value, and taking an absolute value of the second difference value as an absolute size deviation; inputting the quantity difference value and the absolute size deviation into a preset weight function to generate a comprehensive weight factor; And carrying out product operation on the comprehensive weight factors and the initial value of the surface breakage rate to generate the adjusted surface breakage rate.
3. The method of claim 2, wherein the inputting the magnitude difference value and the absolute dimensional deviation into a preset weight function generates a composite weight factor comprising: performing first preset curve processing on the quantity difference value to generate a quantity influence factor; Performing product operation on the absolute size deviation and a preset multiple to generate an amplified size deviation; Performing second preset curve processing on the amplified size deviation to generate a size influence factor; combining the quantity influence factors and the size influence factors according to a preset coupling rule to generate a preliminary combination factor; And inputting the preliminary combination factors into a preset limiting function to generate comprehensive weight factors.
4. A method according to claim 3, wherein said combining the number influencing factors and the size influencing factors according to a preset coupling rule to generate a preliminary combining factor comprises: Performing basic weight conversion on the quantity influence factors to generate basic weight values; performing product operation on the size influence factor and a preset adjustment coefficient to generate a dynamic adjustment factor; Performing product operation on the basic weight value and the dynamic adjustment factor to generate an initial combined value; And adding the initial combination value and a preset compensation factor to generate a preliminary combination factor.
5. The method of claim 1, wherein the performing parallax computation on the binocular vision image data to generate three-dimensional profile data of the preset biomass particles comprises: selecting a plurality of feature points from the left eye image according to the texture features and the edge intersection points of the left eye image; extracting matching points corresponding to the characteristic points from the right eye image to generate a matching point pair set; Calculating the position offset of each matching point in the matching point pair set in the left eye image and the right eye image respectively so as to generate a parallax numerical value set; converting the matching point pair set into a three-dimensional space coordinate set based on the parallax numerical value set and a preset camera distance parameter; connecting adjacent space coordinate points in the three-dimensional space coordinate set to generate a continuous curved surface model; And carrying out surface integrity filling processing on the continuous curved surface model to generate three-dimensional contour data.
6. The method of claim 5, wherein connecting each adjacent spatial coordinate point in the three-dimensional set of spatial coordinates to generate a continuous surface model comprises: constructing a triangular mesh structure by taking each space coordinate point in the three-dimensional space coordinate set as a vertex; performing triangulation processing on the triangular mesh structure to generate an initial triangular mesh; Calculating normal vectors of all triangular faces in the initial triangular mesh to generate a normal vector set; merging adjacent triangular surfaces with consistent directions based on the normal vector set to generate a plurality of smooth curved surface sheets; and connecting boundary edges of the smooth curved surface sheets to generate a continuous curved surface model.
7. The method of claim 1, wherein the performing image recognition processing on the three-dimensional profile data to identify a damaged area of the preset biomass particles comprises: marking curvature extreme points on the three-dimensional contour data to generate a plurality of curvature extreme points; performing region expansion on the three-dimensional contour data based on each curvature extreme point to generate a plurality of candidate regions; calculating average depth data of each candidate region, and comparing each average depth data with a preset depth threshold value respectively to generate a concave region set; calculating the length-width ratio of each candidate region, and comparing each length-width ratio with a preset shape threshold value to generate a crack region set; combining the concave region set and the crack region set to generate an initial damaged region set; and merging adjacent initial damaged areas in the initial damaged area set to generate a damaged area.
8. An image processing-based biomass particle finished product quality grading system, characterized by comprising: the combining module is used for combining left-eye images and right-eye images of different angles of a sampling sample of preset biomass particles to obtain binocular vision image data; The calculation module is used for performing parallax calculation on the binocular vision image data and generating three-dimensional contour data of the preset biomass particles; The identification module is used for carrying out image identification processing on the three-dimensional profile data so as to identify the damage area of the preset biomass particles; The matching module is used for matching the preset damage characteristic with the damage rate associated data and the damage area to obtain a surface damage rate initial value; The adjusting module is used for adjusting the initial value of the surface breakage rate by applying the particle number of the sampling sample and the particle size distribution of the sampling sample to obtain an adjusted surface breakage rate; The dividing module is used for comparing the adjusted surface breakage rate with a preset breakage rate interval so as to divide the quality grade of the finished product of the preset biomass particles according to the comparison result.
9. An electronic device, comprising: A memory for storing a computer program; A processor for implementing the steps of the image processing based biomass particle finished product quality classification method according to any of claims 1 to 7 when executing the computer program.
10. A computer readable storage medium, characterized in that it has stored therein a computer program which, when executed by a processor, is capable of implementing the image processing-based biomass pellet finished product quality classification method according to any one of claims 1 to 7.

Description

Biomass particle finished product quality grading method and system based on image processing Technical Field The application relates to the technical field of biomass particle finished product quality classification, in particular to a biomass particle finished product quality classification method and system based on image processing. Background Along with the wide application of biomass particles as clean energy sources in the fields of industrial boilers, civil heating and power generation, production enterprises are urgently required to carry out efficient and objective quality classification on large batches of particles in a factory inspection link. The damage degree of the particle surface is rapidly and accurately quantified, and comprehensive evaluation is carried out by combining key factors such as particle size distribution, so that the method has become a core requirement for guaranteeing the consistency of the product quality and improving the market competitiveness. The traditional manual detection mode has low efficiency and strong subjectivity, and is difficult to meet the accurate requirement on quality grading in large-scale production, so an automatic detection method capable of rapidly judging the quality of batches through a sampling mode is needed, each batch of products is ensured to meet the national quality standard, and reliable guarantee is provided for the follow-up combustion efficiency and the equipment operation stability. Aiming at the requirement, the existing targeted scheme is a biomass particle quality detection scheme based on monocular image analysis, namely, a standard industrial camera is utilized to shoot a particle sample at a single angle, the color, texture and contour characteristics of the particle surface are extracted through an image processing algorithm, and the area occupation ratio of a defect area is calculated to serve as a quality evaluation index. The system compares the area ratio with a preset threshold value, and directly divides the quality grade of the particles, so as to be used as the basis of batch quality grading. The scheme is relatively simple to operate, can replace manual detection to a certain extent, but has the core limitation of only relying on two-dimensional image information for analysis. The existing scheme has obvious defects that firstly, a two-dimensional information is obtained only by using a single-view image, a three-dimensional structure of particles cannot be accurately restored, so that a large error exists in measurement of a damaged area on the surface of the particles, particularly when the particles are overlapped or shielded, the measurement result is seriously distorted, secondly, the influence of particle size distribution on damage rate evaluation is not considered, the area ratio is directly used as a quality index, so that in batches with uneven particle sizes, the quality grading result is deviated from the actual quality condition, thirdly, only the surface characteristics are focused, key quality indexes such as the overall density and the integrity of the particles are ignored, and the quality grading result is one-sided, so that the actual combustion performance and the use value of the particles cannot be comprehensively reflected. Disclosure of Invention The application aims to provide a biomass particle finished product quality grading method and system based on image processing, which are used for solving the problems that in the prior art, larger errors exist in measurement of particle surface damage areas, deviation exists between quality grading results and actual quality conditions, and the actual combustion performance and use value of particles cannot be comprehensively reflected. In order to solve the technical problems, in a first aspect, the present application provides a biomass particle finished product quality classification method based on image processing, including: Combining left eye images and right eye images of sampling samples of preset biomass particles at different angles to obtain binocular vision image data; parallax calculation is carried out on the binocular vision image data, and three-dimensional contour data of the preset biomass particles are generated; performing image recognition processing on the three-dimensional profile data to recognize a damaged area of the preset biomass particles; matching preset damage characteristics with damage rate associated data and the damage area to obtain a surface damage rate initial value; the initial value of the surface breakage rate is adjusted by using the particle number of the sampling sample and the particle size distribution of the sampling sample, so that the adjusted surface breakage rate is obtained; and comparing the adjusted surface breakage rate with a preset breakage rate interval to divide the quality grade of the finished product of the preset biomass particles according to the comparison result. Optionally, the adjust