CN116883690A - Gray image identification method, device and equipment based on improved multi-scale sampling - Google Patents

Gray image identification method, device and equipment based on improved multi-scale sampling Download PDF

Info

Publication number
CN116883690A
CN116883690A CN202310697241.5A CN202310697241A CN116883690A CN 116883690 A CN116883690 A CN 116883690A CN 202310697241 A CN202310697241 A CN 202310697241A CN 116883690 A CN116883690 A CN 116883690A
Authority
CN
China
Prior art keywords
image
gray
mapping
projection
scale
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN202310697241.5A
Other languages
Chinese (zh)
Inventor
张恩享
朱耿峰
李海建
邓安洲
邓巍
赵勇
陈晓旭
汪臻
李冲
刘腾飞
张轶东
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Huaneng Golmud Photovoltaic Power Generation Co ltd
Xian Thermal Power Research Institute Co Ltd
Original Assignee
Huaneng Golmud Photovoltaic Power Generation Co ltd
Xian Thermal Power Research Institute Co Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Huaneng Golmud Photovoltaic Power Generation Co ltd, Xian Thermal Power Research Institute Co Ltd filed Critical Huaneng Golmud Photovoltaic Power Generation Co ltd
Priority to CN202310697241.5A priority Critical patent/CN116883690A/en
Publication of CN116883690A publication Critical patent/CN116883690A/en
Pending legal-status Critical Current

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/40—Extraction of image or video features
    • G06V10/54—Extraction of image or video features relating to texture
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T5/00—Image enhancement or restoration
    • G06T5/40—Image enhancement or restoration using histogram techniques
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/10—Segmentation; Edge detection
    • G06T7/13—Edge detection
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/40—Analysis of texture
    • G06T7/41—Analysis of texture based on statistical description of texture
    • G06T7/45—Analysis of texture based on statistical description of texture using co-occurrence matrix computation
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/40—Extraction of image or video features
    • G06V10/56—Extraction of image or video features relating to colour
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
    • G06V10/764—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using classification, e.g. of video objects
    • G06V10/765—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using classification, e.g. of video objects using rules for classification or partitioning the feature space
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
    • G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
    • G06V10/80—Fusion, i.e. combining data from various sources at the sensor level, preprocessing level, feature extraction level or classification level
    • G06V10/806—Fusion, i.e. combining data from various sources at the sensor level, preprocessing level, feature extraction level or classification level of extracted features
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02E—REDUCTION OF GREENHOUSE GAS [GHG] EMISSIONS, RELATED TO ENERGY GENERATION, TRANSMISSION OR DISTRIBUTION
    • Y02E10/00—Energy generation through renewable energy sources
    • Y02E10/50—Photovoltaic [PV] energy

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Multimedia (AREA)
  • Computing Systems (AREA)
  • Artificial Intelligence (AREA)
  • Health & Medical Sciences (AREA)
  • Databases & Information Systems (AREA)
  • Evolutionary Computation (AREA)
  • General Health & Medical Sciences (AREA)
  • Medical Informatics (AREA)
  • Software Systems (AREA)
  • Mathematical Physics (AREA)
  • Probability & Statistics with Applications (AREA)
  • Image Analysis (AREA)

Abstract

The invention discloses a gray image identification method, a device and equipment based on improved multi-scale sampling, wherein the identification method comprises the following steps: gray image texture feature extraction, multi-scale sampling processing, projection feature description, gray image three-dimensional space mapping processing and gray image recognition error fitting of fusion mapping results. The invention can extract the texture characteristics of the gray level image and is beneficial to analyzing the image detected by the photovoltaic module EL; the characteristic extraction effect of the image can be improved through multi-scale sampling processing, and the identification accuracy is improved; projection characteristics under different scales can be described through projection characteristic description, so that more useful information can be extracted; the image quality can be improved and the recognition error can be reduced by gray image three-dimensional space mapping processing and gray image recognition error fitting of fusion mapping results; the method can realize effective identification of the gray level image in the detection of the photovoltaic module EL, has high accuracy and has higher application value.

Description

Gray image identification method, device and equipment based on improved multi-scale sampling
Technical Field
The invention belongs to the technical field of image processing, and particularly relates to a gray image identification method, device and equipment based on improved multi-scale sampling.
Background
Image recognition technology is an important content of visual information processing, and research content thereof mainly aims at feature extraction, classification and recognition of images. Image recognition has found widespread use in many fields, such as industrial automation detection, biometric identification, traffic management, and the like.
Electroluminescent (EL) detection of photovoltaic modules is one of the important means of quality detection of photovoltaic products at present. The EL detection is to image and analyze a fluorescent signal emitted by the photovoltaic module under forward bias, so that defects such as microcrack and exclusive area in the photovoltaic module can be effectively detected, and the EL detection plays an important role in ensuring the quality of products. At present, the EL detection mainly adopts manual detection and analysis of EL images, but the EL image quality is lower, the image analysis difficulty is high, the manual detection efficiency is low, misjudgment is easy to generate, and the requirements of large-scale and high-precision detection of photovoltaic products cannot be met.
Gray scale images are the most common type of image used in EL detection, but the difficulty of recognition is high due to the influence of poor image quality and acquisition angle. The existing gray image identification method is mainly based on texture features, shape features, edge features and the like of gray images, is low in identification accuracy and limited in practicability.
The multi-scale image analysis can acquire the characteristics of the image under different scales, so that the robustness of image recognition is enhanced. However, the existing image recognition algorithm based on multi-scale analysis generally needs to set more scale layers, has high computational complexity and poor real-time performance, and is not suitable for EL detection scenes.
In summary, the existing EL detection method and image recognition technology have poor effects and poor real-time performance, and cannot meet the requirement of high-precision detection.
Disclosure of Invention
In order to solve the technical problems in the prior art, the invention aims to provide a gray image identification method, device and equipment based on improved multi-scale sampling, which can effectively identify an EL image, realize automatic and high-precision detection of a photovoltaic module EL and have higher innovation and application value.
In order to achieve the above purpose and achieve the above technical effects, the invention adopts the following technical scheme:
the gray image identification method based on improved multi-scale sampling comprises the following steps:
1. extracting texture features of the gray level image;
2. performing multi-scale sampling treatment;
3. projection feature description;
4. three-dimensional space mapping processing of gray level images;
5. and (5) gray level image recognition error fitting of fusion mapping results.
Further, in the first step, the step of extracting the texture feature of the gray image includes:
firstly, carrying out space mapping projection on an original gray level image to generate an RSP mapping image and an MP mapping image, wherein the RSP mapping image represents ordered gray level value distribution, and the MP mapping image represents ordered pixel point position distribution;
subsequently, feature extraction is performed on the RSP map image and the MP map image, respectively.
Further, in the second step, the step of multi-scale sampling processing includes:
in order to ensure that the image acquired under the multi-scale sampling has the characteristic of projection distribution, when the image is intercepted, the sample dropping processing is carried out once for each row of sub-bands at intervals, and the kth interval of the acquired sub-bands is correspondingly represented as the following calculation formula under the ith scale:
in the method, in the process of the invention,representing the texture feature of the jth sub-band acquired at the ith scale as the corresponding gray image S; u is represented as an original projection interval, W is represented as a rotation angle, a calculation unit is represented as an angle, x and y are respectively represented as mapping projections of an image in a transverse direction and a longitudinal direction, k is represented as a texture interval, and p is represented as a low-frequency subband frequency.
Further, in the third step, the step of projection feature description includes:
assuming that the expression of the image texture in space has a certain characteristic, in order to enhance the effect of texture identification, a projection space with a dimension of MxN is arranged in a sampling space, and a frequency vector is constructed for describing projection characteristics under different dimensions, and the projection characteristics of the gray image under different dimensions are calculated by the following formula:
RIR ′ ＝(SDMV ′ ,FV ′ )(2)
wherein RIR is ′ Expressed as projected features of grey scale images at different scales, SDMUV ′ Represented as frequency vector projection, FV ′ Represented as a supplemental feature;
and integrating all feature information in the projection process into a formula (2), outputting a calculation result as an image texture feature, and finishing effective extraction of the feature.
Further, in the fourth step, the step of gray image three-dimensional space mapping processing includes:
the conventional image processing technology is introduced, the gray value of an image is described by using any one value in 0-256, wherein 256 represents that the image is black, 0 represents that the image is white, the gray value is used as the quality of the image mapped in a three-dimensional space, the high-quality mapping points and edge points of the image are determined according to the distribution of the gray value, the image quality of a three-dimensional mapping area with higher gray value is higher, and the image quality of a three-dimensional mapping area with lower gray value is lower.
Further, in the fifth step, the step of merging the gray image recognition error fitting of the mapping result includes:
because the gray level image is likely to have errors in the identification process due to the gravity center deviation phenomenon after being mapped into the three-dimensional space, in order to reduce the interference of the errors on noise in the identification process, the correction of the identification result is carried out by adopting a fitting error mode, and a gray level density value A is firstly set, wherein the function expression formula of A in the three-dimensional space is as follows:
wherein A is the density distribution of gray values after the image is mapped in a three-dimensional space, M is the information quantity value which can be used for describing the gray values in the image, N is the mean value of a reference image, H is the height of the gray image after the imaging in the space, and C is the visual effect;
and (3) carrying out normalization processing on the corresponding image gray values according to the formula (3) to obtain uniform gray values.
Further, the step of fusing the gray image recognition error fitting of the mapping result further comprises:
correcting the identification image according to the gray level gravity center offset, and reducing errors in the identification process;
and comparing the corrected identification result with an EL image of the actual photovoltaic module, evaluating the accuracy of the identification method, and evaluating and classifying the photovoltaic module according to the identification result.
The invention also discloses a gray image recognition device based on improved multi-scale sampling, which comprises:
the sample acquisition module is used for acquiring a required gray level image;
the preprocessing module is used for preprocessing the gray level image, including image clipping and/or rotation correction;
the gray image texture feature extraction module is used for carrying out space mapping projection on the gray image, generating an RSP mapping image and an MP mapping image, and carrying out feature extraction on the RSP mapping image and the MP mapping image respectively;
the multi-scale sampling processing module is used for obtaining texture characteristics of the gray image under different scales;
the projection characteristic description module is used for obtaining projection characteristics of the gray image under different scales;
the gray image three-dimensional space mapping processing module is used for determining high-quality mapping points and edge points of the image;
a gray image recognition error fitting module;
the result analysis module is used for comparing the corrected identification result with the EL image of the actual photovoltaic module, evaluating the accuracy of the identification method, and evaluating and classifying the photovoltaic module according to the identification result;
the modules work cooperatively to realize effective identification of gray images in detection of the photovoltaic module EL.
Further, the extracted features of the RSP mapping image comprise a gray level histogram and a gray level co-occurrence matrix of the image, and the extracted features of the MP mapping image comprise spatial frequency distribution and spatial position relation.
The invention also discloses a gray image recognition device based on improved multi-scale sampling, which comprises:
a processor for implementing the steps of the gray image recognition method based on improved multi-scale sampling as described above when executing the gray image recognition program stored in the memory.
The invention also discloses a computer readable storage medium having stored thereon a grey scale image recognition program which when executed by a processor implements the steps of the improved multi-scale sampling based grey scale image recognition method as described above.
Compared with the prior art, the invention has the beneficial effects that:
1) Texture features of the gray level image can be extracted, and analysis of the image detected by the photovoltaic module EL is facilitated;
2) The characteristic extraction effect of the image can be improved through multi-scale sampling processing, and the identification accuracy is improved;
3) Projection characteristics under different scales can be described through projection characteristic description, so that more useful information can be extracted;
4) The image quality can be improved and the recognition error can be reduced by gray image three-dimensional space mapping processing and gray image recognition error fitting of fusion mapping results;
5) The accuracy of the identification method can be evaluated by comparing the corrected identification result with the EL image of the actual photovoltaic module, and the photovoltaic module can be evaluated and classified;
6) The method can realize effective identification of gray images in detection of the photovoltaic module EL, realize automatic and high-precision detection of the photovoltaic module EL, improve identification efficiency, and has high accuracy, high reliability, higher innovation and application value.
Drawings
Fig. 1 is a schematic diagram of a gray scale image experimental sample according to embodiment 1 of the present invention;
fig. 2 is a schematic diagram of a sample image recognition result of embodiment 1 of the present invention.
Detailed Description
The present invention is described in detail below so that advantages and features of the present invention can be more easily understood by those skilled in the art, thereby making clear and unambiguous the scope of the present invention.
The following presents a simplified summary of one or more aspects in order to provide a basic understanding of such aspects. This summary is not an extensive overview of all contemplated aspects, and is intended to neither identify key or critical elements of all aspects nor delineate the scope of any or all aspects. Its sole purpose is to present some concepts of one or more aspects in a simplified form as a prelude to the more detailed description that is presented later.
As shown in fig. 1-2, the invention discloses a gray image identification method based on improved multi-scale sampling, which comprises the following steps:
1. gray scale image texture feature extraction
Firstly, in order to meet the gray image recognition requirement, the design of the improved multi-scale sampling algorithm is combined before the algorithm is designed, so that the effective extraction of the texture features of the gray image is carried out. In this process, it should be clear that after the gray-scale image is projected through spatial mapping, the RSP mapping image and the MP mapping image are formed, and therefore, in order to implement extraction of texture features of the gray-scale image, extraction of features of the mapping image should be performed synchronously. It should be noted that, preprocessing may be performed after the gray-scale image is acquired according to actual requirements, including image clipping, rotation correction, and the like. The following is a specific description of generating RSP map image and MP map image by performing spatial map projection on gray level image, and defines each parameter:
in the process of extracting the image texture features, space mapping projection can be carried out on the gray level image, and an RSP mapping image and an MP mapping image are generated. First, the original gray image is projected in a space mapping manner, and each pixel point is mapped into a new coordinate system, which is called a projection space. In the projection space, the mapped pixel points can be ordered according to the gray values, and an RSP mapping image is generated. Meanwhile, the relative position of each pixel point in the projection space can be calculated, and an MP mapping image is generated. The RSP map image here represents the ordered gray value distribution, while the MP map image represents the ordered pixel point location distribution;
among them, the following parameters can be defined:
s. Gray-scale image
RSP ordered gray value distribution (RSP map image)
MP: ordered pixel location distribution (MP map image)
-P projection space
Next, features of the mapped image may be extracted. For RSP mapped images, features that may be extracted include gray level histograms of the images, gray level co-occurrence matrices, and the like. For MP map images, features that may be extracted include spatial frequency distribution, spatial positional relationship, and the like. These features may be used in a subsequent image recognition process.
2. Multi-scale sampling process
In order to ensure that the image acquired under the multi-scale sampling has the characteristic of projection distribution, when the image is intercepted, sample dropping processing is carried out once every one line of sub-bands, and at this time, the kth interval of the acquired sub-bands corresponding to the ith scale can be expressed as the following calculation formula:
in the method, in the process of the invention,representing the image as a corresponding gray image S, and obtaining the texture characteristic of the jth sub-band under the ith scale; u is expressed as an original projection interval; w is expressed as a rotation angle, and the calculation unit is an angle; x and y are respectively expressed as mapping projections of the image in the transverse direction and the longitudinal direction; k is denoted as texture interval; p is expressed as a low frequency subband number。
3. Projection characterization
After the calculation is completed, assuming that the expression of the image texture in space has a certain characteristic, and the effect of identifying the texture is enhanced in the process, a projection space with the dimension of MxN can be set in the sampling space, and a frequency vector is constructed to be used for describing the projection characteristics under different dimensions. The process is shown in the following calculation formula:
RIR ′ ＝(SDMV ′ ,FV ′ )(2)
wherein RIR is ′ Expressed as projected features of the gray scale image at different scales; SDMA ′ Represented as a frequency vector projection; FV (FV) ′ Represented as a supplemental feature. And integrating all feature information in the projection process into a formula, outputting a calculation result, and taking the result as an image texture feature to finish effective feature extraction.
4. Gray scale image three-dimensional space mapping process
The gray value is used as the mapping quality of the image in the three-dimensional space (corresponding to the Z coordinate axis in the three-dimensional space), and the high-quality mapping point and the edge point of the image are determined according to the distribution of the gray value: the conventional image processing technology is introduced, the gray value of the image is described by using any one value from 0 to 256, wherein 256 represents that the image is black, 0 represents that the image is white, the image quality of a three-dimensional mapping area with higher gray value is higher, and the image quality of a three-dimensional mapping area with lower corresponding gray value is lower. If the pixel values of the image are located in the edge region of the image, there will tend to be a high quality image region on the image side. If the image is cut into a rectangular picture, the gravity center point of the region can be used as the image high-quality mapping point, and when the gravity center point of the gray image is offset, the probability of using the pixel value as the image edge point is higher.
5. Gray scale image recognition error fitting fusing mapping results
The gray level image may have errors in the recognition process due to gravity center deviation after being mapped into the three-dimensional space, and in order to reduce the interference of the errors on the noise in the recognition process, the recognition result can be corrected by adopting a fitting error mode. In this process, a gray density value a needs to be set, and the function expression formula of a in the three-dimensional space is as follows:
wherein A is the density distribution of gray values after mapping the image in a three-dimensional space; m is expressed as an information quantity value which can be used for describing gray values in the image; n is expressed as a reference image mean; h is expressed as the height of the gray image after imaging in space; c is expressed as visual effect;
according to the calculation formula (3), carrying out normalization processing on the corresponding image gray values, and obtaining uniform gray values in the mode;
the gray center offset is used as the basis of the identification image, the correction of the identification image is carried out according to the gray center offset, and the error in the identification process is reduced;
and comparing the corrected identification result with an EL image of the actual photovoltaic module, evaluating the accuracy of the identification method, and evaluating and classifying the photovoltaic module according to the identification result.
The invention also discloses a gray image recognition device based on improved multi-scale sampling, which comprises:
the sample acquisition module is used for acquiring a required gray level image;
the preprocessing module is used for preprocessing the gray level image, including image clipping and/or rotation correction;
the gray image texture feature extraction module is used for carrying out space mapping projection on the gray image, generating an RSP mapping image and an MP mapping image, and carrying out feature extraction on the RSP mapping image and the MP mapping image respectively; the extraction features of the RSP mapping image comprise a gray level histogram, a gray level co-occurrence matrix and the like of the image, and the extraction features of the MP mapping image comprise spatial frequency distribution, spatial position relation and the like;
the multi-scale sampling processing module is used for obtaining texture characteristics of the gray image under different scales;
the projection characteristic description module is used for obtaining projection characteristics of the gray image under different scales;
the gray image three-dimensional space mapping processing module is used for determining high-quality mapping points and edge points of the image;
a gray image recognition error fitting module;
the result analysis module is used for comparing the corrected identification result with the EL image of the actual photovoltaic module, evaluating the accuracy of the identification method, and evaluating and classifying the photovoltaic module according to the identification result;
the modules work cooperatively to realize effective identification of gray images in detection of the photovoltaic module EL.
The invention also discloses gray image recognition equipment based on the improved multi-scale sampling, which comprises a processor, wherein the processor is used for realizing the gray image recognition method based on the improved multi-scale sampling when executing the gray image recognition program stored in the memory.
The invention also discloses a computer readable storage medium, the computer readable storage medium stores a gray image recognition program, and the gray image recognition program realizes the steps of the gray image recognition method based on improved multi-scale sampling when being executed by a processor.
Example 1
Before the experiment, gray images in the experiment process are acquired, and in order to ensure the sufficiency of data in the experiment process, the gray images selected in the experiment are all from EL images generated in the detection process of a certain power station. The acquired image samples were 700 images, 420 images of which had defects. If all the images are directly used as experimental sample images, the calculated amount of the experiment is too large, and in order to ensure the reliability of experimental requirements and experimental results, 8 gray-scale images are randomly selected as the sample images of the experiment by combining the use of a visual mechanism. The images need to comprise defect images and defect-free images, after the preliminary acquisition of the images is completed, the terminal equipment is used for cutting gray images, the size of each image is normalized to 256 cm multiplied by 256 cm, and fig. 1 is a gray image experimental sample.
After the acquisition of the image sample in the experimental process is completed, the gray image recognition method based on the improved multi-scale sampling is used for recognizing the sample, so that whether noise exists in the image sample is recognized as a basis for detecting the validity of the design result of the text. When the gray level image is identified to be abnormal, the noise value corresponding to the image is expressed as 1, the defect of the sample image is correspondingly proved, and when the gray level image is identified to be not abnormal, the noise value corresponding to the image is expressed as 0, and the sample is correspondingly proved to be not noisy. The experimental sample is identified according to the identification method of the invention. After recognition, the results are presented on the terminal computer device and plotted as a noise image, as shown in fig. 2. As can be seen from the experimental results shown in fig. 2, the noise values corresponding to the images (1), (2), (3), (4), (5), and (7) in fig. 1 are identified as 1, the noise corresponding to the images (6) and (8) is identified as 0, and the output result is consistent with the actual result.
Thus, the final conclusion of this experiment can be drawn: according to the gray level image recognition method based on improved multi-scale sampling, which is designed by the invention, in practical application, abnormal noise in an image set can be effectively recognized, and image defects can be detected by taking the abnormal noise as a basis.
Parts or structures of the present invention, which are not specifically described, may be existing technologies or existing products, and are not described herein.
The foregoing description is only illustrative of the present invention and is not intended to limit the scope of the invention, and all equivalent structures or equivalent processes or direct or indirect application in other related arts are included in the scope of the present invention.

Claims (10)

1. A gray image recognition method based on improved multi-scale sampling, comprising the steps of:
1. extracting texture features of the gray level image;
2. performing multi-scale sampling treatment;
3. projection feature description;
4. three-dimensional space mapping processing of gray level images;
5. and (5) gray level image recognition error fitting of fusion mapping results.
2. The method for gray image recognition based on improved multiscale sampling according to claim 1, wherein in step one, the step of gray image texture feature extraction comprises:
firstly, carrying out space mapping projection on an original gray level image to generate an RSP mapping image and an MP mapping image, wherein the RSP mapping image represents ordered gray level value distribution, and the MP mapping image represents ordered pixel point position distribution;
subsequently, feature extraction is performed on the RSP map image and the MP map image, respectively.
3. The gray image recognition method based on improved multiscale sampling according to claim 1, wherein in the second step, the step of multiscale sampling processing includes:
in order to ensure that the image acquired under the multi-scale sampling has the characteristic of projection distribution, when the image is intercepted, the sample dropping processing is carried out once for each row of sub-bands at intervals, and the kth interval of the acquired sub-bands is correspondingly represented as the following calculation formula under the ith scale:
in the method, in the process of the invention,representing the texture feature of the jth sub-band acquired at the ith scale as the corresponding gray image S; u is represented as an original projection interval, W is represented as a rotation angle, a calculation unit is represented as an angle, x and y are respectively represented as mapping projections of an image in a transverse direction and a longitudinal direction, k is represented as a texture interval,p is denoted as the low frequency subband number.
4. The method for gray scale image recognition based on improved multiscale sampling of claim 1, wherein in step three, the step of projecting the feature description comprises:
assuming that the expression of the image texture in space has a certain characteristic, in order to enhance the effect of texture identification, a projection space with a dimension of MxN is arranged in a sampling space, and a frequency vector is constructed for describing projection characteristics under different dimensions, and the projection characteristics of the gray image under different dimensions are calculated by the following formula:
RIR ′ ＝(SDMV ′ ,FV ′ )(2)
wherein RIR is ′ Expressed as projected features of grey scale images at different scales, SDMUV ′ Represented as frequency vector projection, FV ′ Represented as a supplemental feature;
and integrating all feature information in the projection process into a formula (2), outputting a calculation result as an image texture feature, and finishing effective extraction of the feature.
5. The gray image recognition method based on improved multiscale sampling according to claim 1, wherein in step four, the step of gray image three-dimensional space mapping processing includes:
and taking the gray value as the mapping quality of the image in the three-dimensional space, and determining high-quality mapping points and edge points of the image according to the distribution of the gray value.
6. The gray image recognition method based on improved multiscale sampling according to claim 1, wherein in step five, the step of fusing the gray image recognition error fit of the mapping result comprises:
setting a gray density value A, wherein the function expression formula of A in the three-dimensional space is as follows:
wherein A is the density distribution of gray values after the image is mapped in a three-dimensional space, M is the information quantity value which can be used for describing the gray values in the image, N is the mean value of a reference image, H is the height of the gray image after the imaging in the space, and C is the visual effect;
and (3) carrying out normalization processing on the corresponding image gray values according to the formula (3) to obtain uniform gray values.
7. The method for gray image recognition based on improved multiscale sampling of claim 6, wherein the step of fusing the gray image recognition error fit of the mapping results further comprises:
correcting the identification image according to the gray level gravity center offset, and reducing errors in the identification process;
and comparing the corrected identification result with an EL image of the actual photovoltaic module, evaluating the accuracy of the identification method, and evaluating and classifying the photovoltaic module according to the identification result.
8. A gray scale image recognition device based on improved multiscale sampling, comprising:
the sample acquisition module is used for acquiring a required gray level image;
the preprocessing module is used for preprocessing the gray level image, including image clipping and/or rotation correction;
the gray image texture feature extraction module is used for carrying out space mapping projection on the gray image, generating an RSP mapping image and an MP mapping image, and carrying out feature extraction on the RSP mapping image and the MP mapping image respectively;
the multi-scale sampling processing module is used for obtaining texture characteristics of the gray image under different scales;
the projection characteristic description module is used for obtaining projection characteristics of the gray image under different scales;
the gray image three-dimensional space mapping processing module is used for determining high-quality mapping points and edge points of the image;
a gray image recognition error fitting module;
the result analysis module is used for comparing the corrected identification result with the EL image of the actual photovoltaic module, evaluating the accuracy of the identification method, and evaluating and classifying the photovoltaic module according to the identification result.
9. A gray scale image recognition device based on improved multiscale sampling, comprising:
a processor for implementing the steps of the improved multiscale sampling based gray image recognition method of any of claims 1-7 when executing a gray image recognition program stored in a memory.
10. A computer readable storage medium, characterized in that the computer readable storage medium has stored thereon a grey-scale image recognition program, which when executed by a processor, implements the steps of the improved multiscale sampling based grey-scale image recognition method according to any of claims 1-7.
CN202310697241.5A 2023-06-13 2023-06-13 Gray image identification method, device and equipment based on improved multi-scale sampling Pending CN116883690A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202310697241.5A CN116883690A (en) 2023-06-13 2023-06-13 Gray image identification method, device and equipment based on improved multi-scale sampling

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202310697241.5A CN116883690A (en) 2023-06-13 2023-06-13 Gray image identification method, device and equipment based on improved multi-scale sampling

Publications (1)

Publication Number Publication Date
CN116883690A true CN116883690A (en) 2023-10-13

Family

ID=88268743

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202310697241.5A Pending CN116883690A (en) 2023-06-13 2023-06-13 Gray image identification method, device and equipment based on improved multi-scale sampling

Country Status (1)

Country Link
CN (1) CN116883690A (en)

Similar Documents

Publication Publication Date Title
CN111951237B (en) Visual appearance detection method
WO2018219054A1 (en) Method, device, and system for license plate recognition
CN107085846B (en) Workpiece surface defect image identification method
CN108229475B (en) Vehicle tracking method, system, computer device and readable storage medium
CN103093201B (en) Vehicle-logo location recognition methods and system
CN111951238A (en) Product defect detection method
CN114594114B (en) Full-automatic online nondestructive testing method for lithium battery cell
CN112308854B (en) Automatic detection method and system for chip surface flaws and electronic equipment
CN101533466B (en) Image processing method for positioning eyes
CN116433666B (en) Board card line defect online identification method, system, electronic equipment and storage medium
CN110110618B (en) SAR target detection method based on PCA and global contrast
CN111126393A (en) Vehicle appearance refitting judgment method and device, computer equipment and storage medium
CN116758045B (en) Surface defect detection method and system for semiconductor light-emitting diode
CN110751619A (en) Insulator defect detection method
CN114049499A (en) Target object detection method, apparatus and storage medium for continuous contour
CN111275040A (en) Positioning method and device, electronic equipment and computer readable storage medium
CN103353941A (en) Natural marker registration method based on viewpoint classification
CN112926592A (en) Trademark retrieval method and device based on improved Fast algorithm
CN112288682A (en) Electric power equipment defect positioning method based on image registration
CN110334775B (en) Unmanned aerial vehicle line fault identification method and device based on width learning
CN110288040B (en) Image similarity judging method and device based on topology verification
CN117557565B (en) Detection method and device for lithium battery pole piece
CN107545565B (en) Solar screen plate detection method
CN110276260B (en) Commodity detection method based on depth camera
CN116843677A (en) Appearance quality detection system and method for sheet metal part

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination