CN114397250A - Wood identification method based on spectral features and image features - Google Patents
Wood identification method based on spectral features and image features Download PDFInfo
- Publication number
- CN114397250A CN114397250A CN202111616170.9A CN202111616170A CN114397250A CN 114397250 A CN114397250 A CN 114397250A CN 202111616170 A CN202111616170 A CN 202111616170A CN 114397250 A CN114397250 A CN 114397250A
- Authority
- CN
- China
- Prior art keywords
- wood
- image
- features
- feature
- training
- 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.)
- Granted
Links
- 239000002023 wood Substances 0.000 title claims abstract description 188
- 230000003595 spectral effect Effects 0.000 title claims abstract description 154
- 238000000034 method Methods 0.000 title claims abstract description 40
- 238000001228 spectrum Methods 0.000 claims abstract description 29
- 238000010606 normalization Methods 0.000 claims abstract description 14
- 238000012545 processing Methods 0.000 claims abstract description 9
- 238000012549 training Methods 0.000 claims description 120
- 238000004422 calculation algorithm Methods 0.000 claims description 49
- 238000007781 pre-processing Methods 0.000 claims description 13
- 238000009826 distribution Methods 0.000 claims description 9
- 239000011159 matrix material Substances 0.000 claims description 8
- 239000000463 material Substances 0.000 claims description 6
- 239000000126 substance Substances 0.000 claims description 6
- 244000137852 Petrea volubilis Species 0.000 claims description 4
- 239000000835 fiber Substances 0.000 claims description 4
- 238000009499 grossing Methods 0.000 claims description 3
- 230000009466 transformation Effects 0.000 claims description 3
- 238000012986 modification Methods 0.000 description 6
- 230000004048 modification Effects 0.000 description 6
- 238000010586 diagram Methods 0.000 description 5
- 239000007787 solid Substances 0.000 description 5
- 238000004590 computer program Methods 0.000 description 4
- 238000003860 storage Methods 0.000 description 4
- 230000003287 optical effect Effects 0.000 description 3
- 244000086363 Pterocarpus indicus Species 0.000 description 2
- 235000009984 Pterocarpus indicus Nutrition 0.000 description 2
- 238000013145 classification model Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 239000004973 liquid crystal related substance Substances 0.000 description 2
- 241000894007 species Species 0.000 description 2
- 238000012706 support-vector machine Methods 0.000 description 2
- 235000007173 Abies balsamea Nutrition 0.000 description 1
- 239000004857 Balsam Substances 0.000 description 1
- 244000068988 Glycine max Species 0.000 description 1
- 235000010469 Glycine max Nutrition 0.000 description 1
- 244000018716 Impatiens biflora Species 0.000 description 1
- 241000522650 Myroxylon Species 0.000 description 1
- 241000522660 Pericopsis Species 0.000 description 1
- 244000100205 Robinia Species 0.000 description 1
- 241000219784 Sophora Species 0.000 description 1
- 210000003484 anatomy Anatomy 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000005520 cutting process Methods 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 230000002349 favourable effect Effects 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000005498 polishing Methods 0.000 description 1
- 239000002994 raw material Substances 0.000 description 1
- 230000000717 retained effect Effects 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
- G01N21/17—Systems in which incident light is modified in accordance with the properties of the material investigated
- G01N21/25—Colour; Spectral properties, i.e. comparison of effect of material on the light at two or more different wavelengths or wavelength bands
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
- G01N21/17—Systems in which incident light is modified in accordance with the properties of the material investigated
- G01N21/25—Colour; Spectral properties, i.e. comparison of effect of material on the light at two or more different wavelengths or wavelength bands
- G01N21/31—Investigating relative effect of material at wavelengths characteristic of specific elements or molecules, e.g. atomic absorption spectrometry
- G01N21/35—Investigating relative effect of material at wavelengths characteristic of specific elements or molecules, e.g. atomic absorption spectrometry using infrared light
- G01N21/359—Investigating relative effect of material at wavelengths characteristic of specific elements or molecules, e.g. atomic absorption spectrometry using infrared light using near infrared light
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/40—Analysis of texture
-
- 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
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/30—Computing systems specially adapted for manufacturing
Landscapes
- Physics & Mathematics (AREA)
- Spectroscopy & Molecular Physics (AREA)
- General Physics & Mathematics (AREA)
- Biochemistry (AREA)
- Chemical & Material Sciences (AREA)
- Analytical Chemistry (AREA)
- Life Sciences & Earth Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Immunology (AREA)
- Pathology (AREA)
- Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Theoretical Computer Science (AREA)
- Image Analysis (AREA)
Abstract
The invention relates to a wood identification method based on spectral features and image features, which comprises the following steps: s1, acquiring spectral data and image data of the wood to be identified; s2, extracting three groups of feature sets aiming at the spectrum data and the image data; each set of feature sets comprises a plurality of predetermined spectral features and a plurality of predetermined image features corresponding to the set of feature sets; s3, combining a plurality of predetermined spectral features and a plurality of image features in each group of feature sets, and respectively obtaining final features corresponding to each feature set; s4, respectively carrying out normalization processing on the three final features, inputting the final features after the normalization processing into a trained classifier, and respectively obtaining three corresponding classification results; and S5, determining the final recognition result of the wood based on the three classification results.
Description
Technical Field
The invention relates to the technical field of wood identification, in particular to a wood identification method based on spectral features and image features.
Background
The correct identification of wood species is the first step of resisting illegal wood cutting trade and scientific and reasonable processing and utilization of wood raw materials, and the traditional wood identification method established on the basis of the wood anatomical structure needs to be operated by skilled professionals, and the test period is long and the requirement of on-site quick wood identification is difficult to meet. Emerging technologies (such as DNA barcodes) can realize accurate identification of wood species, but are limited by expensive detection cost and professional operation under laboratory conditions, and are not favorable for popularization and application in actual production activities.
However, the existing method for identifying wood only by using the spectral data of wood needs to ensure the identification accuracy of wood by increasing the number and quality of spectral databases, but the process is not easy to be realized in a short period.
Disclosure of Invention
Technical problem to be solved
In view of the above disadvantages and shortcomings of the prior art, the present invention provides a method for wood identification based on spectral features and image features, which solves the technical problem of limited wood identification accuracy under the condition of limited spectral database by adding image feature data.
(II) technical scheme
In order to achieve the purpose, the invention adopts the main technical scheme that:
the embodiment of the invention provides a wood identification method based on spectral features and image features, which comprises the following steps:
s1, acquiring spectral data and image data of the wood to be identified;
s2, extracting three groups of feature sets aiming at the spectrum data and the image data;
each set of feature sets comprises a plurality of predetermined spectral features and a plurality of predetermined image features corresponding to the set of feature sets;
s3, combining a plurality of predetermined spectral features and a plurality of image features in each group of feature sets, and respectively obtaining final features corresponding to each feature set;
s4, respectively carrying out normalization processing on the three final features, inputting the final features after the normalization processing into a trained classifier, and respectively obtaining three corresponding classification results;
the trained classifier is obtained by training the classifier through the final feature for training;
the final features for training include: the method comprises the steps that spectrum data used for training and extracted from image data used for training, corresponding to each type of wood in multiple types of wood, and three groups of final features which are extracted from the image data used for training and are combined with feature sets used for training and corresponding to the type of wood;
and S5, determining the final recognition result of the wood based on the three classification results.
Preferably, the S3 specifically includes:
combining a plurality of predetermined spectral features and a plurality of image features in each group of feature sets by adopting a formula (1), and respectively obtaining final features corresponding to each feature set, wherein the formula (1) is as follows:
f1 is the final feature corresponding to the feature set;
[yn]a set of a plurality of predetermined spectral features corresponding to the feature set;
[wm]a set of a plurality of predetermined image features corresponding to the feature set;
[yn，wm]a plurality of predetermined spectral features and a plurality of predetermined image features corresponding to the feature set;
wherein A is1Is a first predetermined coefficient;
A2a second predetermined coefficient;
B1a third predetermined coefficient;
B2is a fourth predetermined coefficient.
Preferably, the S2 includes:
s21, preprocessing the spectral data to obtain preprocessed spectral data;
the preprocessing operation sequentially comprises one or more of smoothing operation, first-order differential operation or second-order differential operation, standard variable transformation operation and centralization operation;
and S22, extracting three groups of feature sets aiming at the image data and the spectrum data after the preprocessing operation.
Preferably, the first and second liquid crystal materials are,
the wavelength range in the spectral data comprises visible light and near infrared spectral bands;
the image data is an image scanned in equal proportion to the wood surface to be identified corresponding to the spectral data during collection; the resolution of the image data is greater than 512 x 512 pixels;
the wood surface to be identified is previously ground by sand paper with 400, 800 and 1000 meshes in sequence, and the wood surface is polished by a fiber wheel after being ground.
Preferably, before S1, the method further includes:
a1, acquiring a training set of spectral data and image data;
the training set of the spectral data and the image data comprises spectral data and image data, wherein the spectral data and the image data are used for training and correspond to each type of wood in multiple types of wood;
a2, determining each feature set in three groups of feature sets corresponding to each type of wood in the multiple types of wood according to the training set of the spectrum data and the image data;
a3, respectively acquiring three final features for training corresponding to each type of wood based on each feature set in the three sets of feature sets corresponding to each type of wood;
a4, normalizing the three final features for training corresponding to each type of wood in the multiple types of wood to obtain the final features for training after the normalization, inputting the final features for training after the normalization into a classifier, and training by the classifier until the accuracy of the classifier meets a preset value to finish the training to obtain the trained classifier.
Preferably, the a2 includes:
a21, preprocessing each spectral data for training in the set of spectral data for training in the training set of the spectral data and the image data to obtain preprocessed spectral data for training;
a22, aiming at image data which are used for training and correspond to any type of wood in multiple types of wood, adopting a gray level co-occurrence matrix method to extract Z image features which are used for training and correspond to the type of wood;
a23, respectively acquiring the distribution range of each image feature in the image data for training corresponding to each type of wood by adopting a formula (2) according to the image data for training corresponding to each type of wood in the multiple types of wood and the multiple image features;
the formula (2):
wherein，
W for the R-th wood of multiple types of woodiImage feature distribution range;
wi，minis wiA minimum value in the image feature data;
wi，maxis wiA maximum value in the image feature data;
n is the number of types of wood included in the training set of spectral and image data;
wiis the ith image feature in the plurality of image features;
a24, respectively acquiring the intersection of each image feature in the wood of multiple types by adopting a formula (3) based on the distribution range of each image feature in the image data which is corresponding to each type of wood and used for training;
the formula (3) is:
win is w of N types of woodiIntersection of image feature data;
wi，max Nw for N types of woodiA minimum value in an intersection of image feature data;
wi，max Nw for N types of woodiA maximum value in an intersection of image feature data;
a25, acquiring image feature difference coefficients of the wood in the various types by adopting a formula (4) based on the intersection of each image feature in the wood in the various types;
the formula (4) is:
wherein Z is the number of image features; cNA difference coefficient of image characteristics for N types of wood;
and A26, determining each feature in the three sets of feature sets corresponding to each type of wood in the multiple types of wood based on the difference coefficient.
Preferably, the a26 includes:
a261, if the difference coefficient, satisfies 0.8 < CNIf the spectrum characteristic is less than or equal to 1, aiming at the preprocessed spectrum data set for training, respectively determining the spectrum characteristics corresponding to the preset threshold T in the Relief-F algorithm when the threshold T is 0.25, 0.55 and 0.75 by adopting the Relief-F algorithm;
if the difference coefficient satisfies 0 < CNIf the sum is less than or equal to 0.6, respectively determining spectral characteristics corresponding to preset threshold values T in a Relief-F algorithm when the threshold values T are 0.35, 0.65 and 0.85 by adopting the Relief-F algorithm according to the preprocessed spectral data set sum for training;
if the difference coefficient satisfies 0.6 < CNIf the spectrum characteristic is less than or equal to 0.8, respectively determining the spectrum characteristics respectively corresponding to the preset threshold T in the Relief-F algorithm when the threshold T is 0.45, 0.8 and 0.95 by adopting the Relief-F algorithm aiming at the preprocessed spectrum data set for training;
a262, calculating a plurality of image features in image data used for training by adopting a Pearson correlation coefficient algorithm, and acquiring a correlation coefficient p among the image features;
a263, if the difference coefficient, satisfy 0.8 < CNIf the absolute value of the correlation coefficient p is less than or equal to 1, the absolute values of the correlation coefficients p are respectively determined<0.25 image feature, absolute value of correlation coefficient p<0.55 image feature, absolute value of correlation coefficient p<An image feature of 0.75;
if the difference coefficient satisfies 0 < CNLess than or equal to 0.6, determining the absolute value of the correlation coefficient p<0.35 image feature, absolute value of correlation coefficient p<0.65 image feature, absolute value of correlation coefficient p<An image characteristic of 0.85;
if the difference coefficient satisfies 0.6 < CNLess than or equal to 0.8, determining the absolute of the correlation coefficient pValue of<0.45 image feature, absolute value of correlation coefficient p<0.8 image feature, absolute value of correlation coefficient p<An image characteristic of 0.95;
a264, if the difference coefficient, satisfies 0.8 < CNDetermining three groups of feature sets corresponding to each type of wood in the multiple types of wood when the number of the wood is less than or equal to 1;
wherein, three sets of feature sets corresponding to each type of wood in the multiple types of wood comprise: a first group of feature sets, a second group of feature sets, and a third group of feature sets;
the first set of feature sets includes: when a preset threshold value T in a Relief-F algorithm is 0.25, corresponding spectral characteristics and image characteristics with the absolute value of the correlation coefficient p being less than 0.25 are obtained;
the second set of feature sets includes: when a preset threshold value T in a Relief-F algorithm is 0.55, corresponding spectral characteristics and image characteristics with the absolute value of the correlation coefficient p being less than 0.55 are obtained;
the third set of feature sets comprises: when a preset threshold value T in a Relief-F algorithm is 0.75, corresponding spectral characteristics and image characteristics with the absolute value of the correlation coefficient p being less than 0.75 are obtained;
if the difference coefficient satisfies 0 < CNDetermining three groups of feature sets corresponding to each type of wood in the multiple types of wood, wherein the three groups of feature sets are less than or equal to 0.6;
wherein, three sets of feature sets corresponding to each type of wood in the multiple types of wood comprise: a fourth group of feature sets, a fifth group of feature sets and a sixth group of feature sets;
the fourth set of features comprises: when a preset threshold value T in a Relief-F algorithm is 0.35, corresponding spectral characteristics and image characteristics with the absolute value of the correlation coefficient p being less than 0.35 are obtained;
the fifth set of features comprises: when a preset threshold value T in a Relief-F algorithm is 0.65, corresponding spectral characteristics and image characteristics with the absolute value of the correlation coefficient p being less than 0.65 are obtained;
the sixth set of feature sets comprises: when the value of T in a Relief-F algorithm is 0.85, corresponding spectral characteristics and image characteristics with the absolute value of the correlation coefficient p being less than 0.85 are obtained;
if the difference coefficient satisfies 0.6 < CNDetermining three groups of feature sets corresponding to each type of wood in the multiple types of wood, wherein the three groups of feature sets are less than or equal to 0.8;
wherein, three sets of feature sets corresponding to each type of wood in the multiple types of wood comprise: a seventh group of feature sets, an eighth group of feature sets, and a ninth group of feature sets;
the seventh set of features comprises: when a preset threshold value T in a Relief-F algorithm is 0.45, corresponding spectral characteristics and image characteristics with the absolute value of the correlation coefficient p being less than 0.45 are obtained;
the eighth set of features comprises: when a preset threshold value T in a Relief-F algorithm is 0.8, corresponding spectral characteristics and image characteristics with the absolute value of the correlation coefficient p being less than 0.8 are obtained;
the ninth set of features includes: when a preset threshold value T in a Relief-F algorithm is taken as a value of 0.95, the corresponding spectral characteristics and the image characteristics with the absolute value of the correlation coefficient p being less than 0.95 are obtained.
Preferably, a3 specifically includes:
aiming at each group of feature set corresponding to each type of wood, adopting a formula (5) to obtain corresponding final features for training;
the formula (5) is:
f2 is the final feature for training corresponding to the set of feature sets;
[ys]all spectral feature sets used for training in the set of feature sets;
[wg]all image feature sets used for training in the set of feature sets;
[ys，wg]for all spectral features used for training and image features used for training in the set of feature sets;
Preferably, the first and second liquid crystal materials are,
the three groups of feature sets extracted for the spectral data and the image data in the S2 satisfy 0.8 < C in the difference coefficientNUnder the condition of less than or equal to 1, the following are respectively: a first group of feature sets, a second group of feature sets, and a third group of feature sets;
the three sets of feature sets extracted for the spectral data and the image data in the S2 satisfy 0 < C in the difference coefficientNUnder the condition of less than or equal to 0.6, the following are respectively: a fourth group of feature sets, a fifth group of feature sets and a sixth group of feature sets;
the three groups of feature sets extracted for the spectral data and the image data in the S2 satisfy 0.6 < C in the difference coefficientNUnder the condition of less than or equal to 0.8, the following are respectively: a seventh set of features, an eighth set of features, and a ninth set of features.
Preferably, the S5 specifically includes: and judging whether the three classification results are the same, and if at least two classification results in the three classification results are the same, taking the same at least two classification results as final identification results.
(III) advantageous effects
The invention has the beneficial effects that: according to the wood identification method based on the spectral features and the image features, the final features obtained by combining the spectral features and the image features of the wood to be detected are input into the trained classifier for classification and identification, and compared with the prior art, the method can extract the features from the spectral data, so that the chemical, optical and other information of the wood sample is retained to the maximum extent, and meanwhile, high-dimensional data is converted into low-dimensional data. By combining the spectral characteristics with the image characteristics, important information such as chemical components and optics contained in the spectrum can be combined with the structural information of the wood surface, so that final characteristics which are easier to identify by a classifier are obtained, and the accuracy of classifier identification is improved.
Drawings
Fig. 1 is a flowchart of a wood identification method based on spectral features and image features according to the present invention.
Detailed Description
For the purpose of better explaining the present invention and to facilitate understanding, the present invention will be described in detail by way of specific embodiments with reference to the accompanying drawings.
In order to better understand the above technical solutions, exemplary embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. While exemplary embodiments of the invention are shown in the drawings, it should be understood that the invention can be embodied in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the invention to those skilled in the art.
Referring to fig. 1, the present embodiment provides a wood identification method based on spectral features and image features, including:
and S1, acquiring the spectral data and the image data of the wood to be identified.
And S2, extracting three groups of feature sets aiming at the spectral data and the image data.
Each set of feature sets includes a predetermined plurality of spectral features and a plurality of image features corresponding to the set of feature sets.
For example, the three sets of feature sets in this embodiment have different features, for example, there are 5 predetermined spectral features and 5 predetermined image features in the first set of feature sets. There are 7 predetermined spectral features and 7 predetermined image features in the second set of features. There are 10 predetermined spectral features and 10 predetermined image features in the third set of features.
And S3, combining a plurality of predetermined spectral features and a plurality of image features in each group of feature sets, and respectively acquiring final features corresponding to each feature set.
And S4, respectively carrying out normalization processing on the three final features, inputting the final features after the normalization processing into a trained classifier, and respectively obtaining three corresponding classification results.
The trained classifier is obtained after the classifier is trained by the final feature used for training.
The middle classifier in this embodiment is a support vector machine model or a neural network classification model, and this embodiment is not limited in particular. The classifier obtained after training in this embodiment has a correct rate of more than 90%.
The final features for training include: the final characteristics are combined by the spectrum data used for training corresponding to each type of wood in the wood with the three groups of characteristic sets used for training, which are extracted as the image data used for training and correspond to the wood.
And S5, determining the final recognition result of the wood based on the three classification results.
In practical application of this embodiment, the S3 specifically includes:
combining a plurality of predetermined spectral features and a plurality of image features in each group of feature sets by adopting a formula (1), and respectively obtaining final features corresponding to each feature set, wherein the formula (1) is as follows:
f1 is the final feature corresponding to the feature set.
[yn]Is a set of a plurality of predetermined spectral features corresponding to the feature set.
[wm]Is a predetermined set of a plurality of image features corresponding to the feature set.
[yn，wm]A set of a plurality of spectral features and a plurality of image features that are predetermined for the set of features.
Wherein A is1Is a first predetermined coefficient.
A2Is a second predetermined coefficient.
B1Is a third predetermined coefficient.
B2Is a fourth predetermined coefficient.
In practical applications of this embodiment, the S2 includes:
and S21, performing preprocessing operation on the spectral data to obtain the spectral data after the preprocessing operation.
The preprocessing operation sequentially comprises one or more of smoothing operation, first-order differential operation or second-order differential operation, standard variable transformation operation and centralization operation.
And S22, extracting three groups of feature sets aiming at the image data and the spectrum data after the preprocessing operation.
In practical applications of this embodiment, the wavelength range in the spectral data includes visible and near infrared spectral bands.
The image data is an image scanned in equal proportion to the wood surface to be identified corresponding to the spectral data during collection; the resolution of the image data is greater than 512 x 512 pixels.
The wood surface to be identified is previously ground by sand paper with 400, 800 and 1000 meshes in sequence, and the wood surface is polished by a fiber wheel after being ground.
In practical application of this embodiment, before S1, the method further includes:
a1, acquiring a training set of spectral data and image data.
The training set of the spectral data and the image data comprises spectral data and image data for training, which correspond to each type of wood in the multiple types of wood.
A2, determining each feature set in three groups of feature sets corresponding to each type of wood in the multiple types of wood according to the training set of the spectrum data and the image data.
A3, respectively obtaining three final features for training corresponding to each type of wood based on each feature set in the three sets of feature sets corresponding to each type of wood.
A4, normalizing the three final features for training corresponding to each type of wood in the multiple types of wood to obtain the final features for training after the normalization, inputting the final features for training after the normalization into a classifier, and training by the classifier until the accuracy of the classifier meets a preset value to finish the training to obtain the trained classifier.
In practical application of this embodiment, the a2 includes:
and A21, preprocessing is carried out on each spectral data for training in the set of spectral data for training in the training set of the spectral data and the image data, and preprocessed spectral data for training are obtained.
And A22, adopting a gray level co-occurrence matrix method to extract Z image features in the image data for training corresponding to any type of wood in the wood types.
Specifically, the texture feature extraction by using the gray level co-occurrence matrix method specifically includes: the gray level co-occurrence matrix counts the probability of the pixel points of a gray level value j and a gray level value k appearing in the distance d and different directions theta on the image (the gray level value is h).
The expression is as follows: q (j, k, d, θ) { [ (x, y), (x + a, y + b) | f (x, y) ═ j; f (x + a, y + b) ═ k ] }.
(x, y) is an arbitrary point in the image.
(x + a, y + b) is another point of deviation from (x, y).
f (x, y) represents the gray value at point (x, y).
f (x + a, y + b) represents the grayscale value at point (x + a, y + b).
d is the distance.
θ is the direction, taken in this example as 0 °, 45 °, 90 °, and 135 °.
In this embodiment, 14 (in this case, Z is 14) image features are extracted by the gray level co-occurrence matrix as follows:
Wherein:
w1，...，w14are 14 image features extracted by the gray level co-occurrence matrix.
q (j, k, d, θ) is a gray level co-occurrence matrix calculation expression.
h is the gray value of the input image.
The gray values of two different point pixels on the j, k image.
u1、u2Is the mean value.
d1、d2Is the standard deviation.
A23, respectively obtaining the distribution range of each image feature in the image data for training corresponding to each type of wood according to the image data for training corresponding to each type of wood and the plurality of image features by adopting a formula (2).
The formula (2):
wherein the content of the first and second substances,
w for the R-th wood of multiple types of woodiImage feature distribution range.
wi，minIs wiA minimum value in the image characteristic data.
W′i，maxIs wiA maximum value in the image characteristic data.
N is the number of types of wood included in the training set of spectral and image data.
wiIs the ith image feature of the plurality of image features.
A24, based on the distribution range of each image feature in the image data for training corresponding to each type of wood, using formula (3), respectively obtaining the intersection of each image feature in the wood of the plurality of types.
The formula (3) is:
wi Nw for N types of woodiIntersection of image feature data.
wi，min NW for N types of woodiThe minimum value in the intersection of image feature data.
wi，max NW for N types of woodiMaximum value in the intersection of image feature data.
A25, based on the intersection of each image feature in the wood of multiple types, adopting formula (4) to obtain the image feature difference coefficient of the wood of multiple types.
The formula (4) is:
wherein Z is the number of image features; cNIs the coefficient of difference in image characteristics of N types of wood.
And A26, determining each feature in the three sets of feature sets corresponding to each type of wood in the multiple types of wood based on the difference coefficient.
In practical application of this embodiment, the a26 includes:
and A261, if the difference coefficient satisfies 0.8 < CN < 1, adopting a Relief-F algorithm for the preprocessed spectral data set for training, and respectively determining the spectral features respectively corresponding to the preset threshold T in the Relief-F algorithm when the threshold T is 0.25, 0.55 and 0.75.
In the embodiment, the Relief-F algorithm presets a threshold T, and then selects the corresponding spectral feature having the relevant statistical component greater than the threshold T from the N types of preprocessed spectral data sets.
If the difference coefficient satisfies 0 < CNAnd if the sum is less than or equal to 0.6, respectively determining the spectral characteristics corresponding to the preset threshold T in the Relief-F algorithm when the threshold T is 0.35, 0.65 and 0.85 by adopting the Relief-F algorithm according to the preprocessed spectral data set sum for training.
If the difference coefficient satisfies 0.6 < CNAnd if the spectrum characteristic is less than or equal to 0.8, adopting a Relief-F algorithm for the preprocessed spectrum data set for training, and respectively determining the spectrum characteristics corresponding to the preset threshold T in the Relief-F algorithm when the threshold T is 0.45, 0.8 and 0.95.
And A262, calculating a plurality of image features in the image data for training by adopting a Pearson correlation coefficient algorithm, and acquiring a correlation coefficient p among the plurality of image features.
A263, if the difference coefficient, satisfy 0.8 < CNIf the absolute value of the correlation coefficient p is less than or equal to 1, the absolute values of the correlation coefficients p are respectively determined<0.25 image feature, absolute value of correlation coefficient p<0.55 image feature, absolute value of correlation coefficient p<Image feature of 0.75.
If the difference coefficient satisfies 0 < CNLess than or equal to 0.6, determining the absolute value of the correlation coefficient p<0.35 image feature, absolute value of correlation coefficient p<0.65 image feature, absolute value of correlation coefficient p<Image feature of 0.85.
If the difference coefficient satisfies 0.6 < CNLess than or equal to 0.8, determining the absolute value of the correlation coefficient p<0.45Image feature of (1), absolute value of correlation coefficient p<0.8 image feature, absolute value of correlation coefficient p<Image feature of 0.95.
A264, if the difference coefficient, satisfies 0.8 < CNAnd determining three sets of feature sets corresponding to each type of wood in the multiple types of wood according to the condition that the wood is not more than 1.
Wherein, three sets of feature sets corresponding to each type of wood in the multiple types of wood comprise: a first set of features, a second set of features, and a third set of features.
The first set of feature sets includes: when a preset threshold value T in a Relief-F algorithm is 0.25, the corresponding spectral characteristics and the image characteristics of which the absolute value of the correlation coefficient p is less than 0.25 are obtained.
The second set of feature sets includes: when a preset threshold value T in a Relief-F algorithm is 0.55, the corresponding spectral characteristics and the image characteristics of which the absolute value of the correlation coefficient p is less than 0.55 are obtained.
The third set of feature sets comprises: when a preset threshold value T in the Relief-F algorithm is 0.75, the corresponding spectral characteristics and the image characteristics of which the absolute value of the correlation coefficient p is less than 0.75 are obtained.
If the difference coefficient satisfies 0 < CNAnd determining three sets of feature sets corresponding to each type of wood in the multiple types of wood according to the condition that the number of the feature sets is less than or equal to 0.6.
Wherein, three sets of feature sets corresponding to each type of wood in the multiple types of wood comprise: a fourth group of feature sets, a fifth group of feature sets, and a sixth group of feature sets.
The fourth set of features comprises: when a preset threshold value T in the Relief-F algorithm is 0.35, the corresponding spectral characteristics and the image characteristics of which the absolute value of the correlation coefficient p is less than 0.35 are obtained.
The fifth set of features comprises: when a preset threshold value T in a Relief-F algorithm is 0.65, the corresponding spectral characteristics and the image characteristics of which the absolute value of the correlation coefficient p is less than 0.65 are obtained.
The sixth set of feature sets comprises: and when the value of T in the Relief-F algorithm is 0.85, the corresponding spectral characteristics and the image characteristics of which the absolute value of the correlation coefficient p is less than 0.85 are obtained.
If the difference coefficient satisfies 0.6 < CNAnd determining three sets of feature sets corresponding to each type of wood in the multiple types of wood according to the condition that the number of the feature sets is less than or equal to 0.8.
Wherein, three sets of feature sets corresponding to each type of wood in the multiple types of wood comprise: a seventh set of features, an eighth set of features, and a ninth set of features.
The seventh set of features comprises: when a preset threshold value T in a Relief-F algorithm is 0.45, the corresponding spectral characteristics and the image characteristics of which the absolute value of the correlation coefficient p is less than 0.45 are obtained.
The eighth set of features comprises: when a preset threshold value T in a Relief-F algorithm is 0.8, the corresponding spectral characteristics and the image characteristics of which the absolute value of the correlation coefficient p is less than 0.8 are obtained.
The ninth set of features includes: when a preset threshold value T in a Relief-F algorithm is taken as a value of 0.95, the corresponding spectral characteristics and the image characteristics with the absolute value of the correlation coefficient p being less than 0.95 are obtained.
In practical application of this embodiment, the a3 specifically includes:
and (5) acquiring corresponding final features for training by adopting a formula (5) for each group of feature set corresponding to each type of wood.
The formula (5) is:
f2 is the final feature for training corresponding to the set of feature sets.
[ys]For all spectral feature sets used for training in the set of feature sets.
[wg]The set of image features for training is all in the set of feature sets.
[ys，wg]For all spectral features used for training and image features used for training in the set of feature sets.
In practical application of the present embodiment, the three sets of feature sets extracted for the spectral data and the image data in S2 satisfy 0.8 < C in the difference coefficientNUnder the condition of less than or equal to 1, the following are respectively: a first set of features, a second set of features, and a third set of features.
The three sets of feature sets extracted for the spectral data and the image data in the S2 satisfy 0 < C in the difference coefficientNUnder the condition of less than or equal to 0.6, the following are respectively: a fourth group of feature sets, a fifth group of feature sets, and a sixth group of feature sets.
The three groups of feature sets extracted for the spectral data and the image data in the S2 satisfy 0.6 < C in the difference coefficientNUnder the condition of less than or equal to 0.8, the following are respectively: a seventh set of features, an eighth set of features, and a ninth set of features.
In practical application of this embodiment, the S5 specifically includes: and judging whether the three classification results are the same, and if at least two classification results in the three classification results are the same, taking the same at least two classification results as final identification results.
Authentication
The wood identification method based on the spectral characteristics and the image characteristics in the embodiment is adopted to respectively identify 6 kinds of common pteridellus solid wood floor woods in the floor market (balsam pea (Myroxylon sp.); locust (Sophora sp.); locust (Robinia sp.); big American soybean (Pericopsis sp.); dipterox sp.; rosewood (Pterocarpus sp.)), and before identification, 6 kinds of woods to be identified are subjected to grinding and polishing treatment, and the surfaces of the solid wood floor wood sample are ground by using sand paper with 400, 800 and 1000 meshes in sequence and polished by using a fiber wheel. Then, the spectral data of the 6 wood samples of the solid wood floor to be identified are collected, and a Field Spec near infrared spectrometer of ASD company is used for collecting the spectral data, wherein the spectral wavelength range is 350-2500 nm. Then, acquiring image data of the 6 solid wood floor wood samples to be identified, and scanning the original size image data of the wood surface by using a scanner corresponding to the surface of the sample acquired by the spectrum, wherein the image resolution is 512 multiplied by 512 pixels, and the data format storage format is JPG.
Then, the spectral features and the image features are combined according to the wood identification method based on the spectral features and the image features in the embodiment to generate corresponding final features, and then the final features corresponding to the 6 solid wood floor wood samples to be identified are respectively input into a classifier to obtain a final identification result. The following table shows the accuracy of classification and discrimination of the 6 tree sample data by a classifier (support vector machine classification model).
In the wood identification method based on the spectral features and the image features, the final features obtained by combining the spectral features and the image features of the wood to be detected are input into the trained classifier for classification and identification, and compared with the prior art, the method can extract the features from the spectral data, and convert high-dimensional data into low-dimensional data while maximally retaining chemical, optical and other information of the wood sample. By combining the spectral characteristics with the image characteristics, important information such as chemical components and optics contained in the spectrum can be combined with the structural information of the wood surface, so that final characteristics which are easier to identify by a classifier are obtained, and the accuracy of classifier identification is improved.
Since the system described in the above embodiment of the present invention is a system used for implementing the method of the above embodiment of the present invention, a person skilled in the art can understand the specific structure and the modification of the system based on the method described in the above embodiment of the present invention, and thus the detailed description is omitted here. All systems adopted by the method of the above embodiments of the present invention are within the intended scope of the present invention.
As will be appreciated by one skilled in the art, embodiments of the present invention may be provided as a method, system, or computer program product. Accordingly, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The present invention is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the invention. It will be understood that each flow and/or block of the flow diagrams and/or block diagrams, and combinations of flows and/or blocks in the flow diagrams and/or block diagrams, can be implemented by computer program instructions.
It should be noted that in the claims, any reference signs placed between parentheses shall not be construed as limiting the claim. The word "comprising" does not exclude the presence of elements or steps not listed in a claim. The word "a" or "an" preceding an element does not exclude the presence of a plurality of such elements. The invention may be implemented by means of hardware comprising several distinct elements, and by means of a suitably programmed computer. In the claims enumerating several means, several of these means may be embodied by one and the same item of hardware. The use of the terms first, second, third and the like are for convenience only and do not denote any order. These words are to be understood as part of the name of the component.
Furthermore, it should be noted that in the description of the present specification, the description of the term "one embodiment", "some embodiments", "examples", "specific examples" or "some examples", etc., means that a specific feature, structure, material or characteristic described in connection with the embodiment or example is included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the terms used above are not necessarily intended to refer to the same embodiment or example. Furthermore, the particular features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples. Furthermore, various embodiments or examples and features of different embodiments or examples described in this specification can be combined and combined by one skilled in the art without contradiction.
While preferred embodiments of the present invention have been described, additional variations and modifications in those embodiments may occur to those skilled in the art once they learn of the basic inventive concepts. Therefore, the claims should be construed to include preferred embodiments and all changes and modifications that fall within the scope of the invention.
It will be apparent to those skilled in the art that various modifications and variations can be made in the present invention without departing from the spirit or scope of the invention. Thus, if such modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention should also include such modifications and variations.
Claims (10)
1. A method for identifying wood based on spectral and image features, comprising:
s1, acquiring spectral data and image data of the wood to be identified;
s2, extracting three groups of feature sets aiming at the spectrum data and the image data;
each set of feature sets comprises a plurality of predetermined spectral features and a plurality of predetermined image features corresponding to the set of feature sets;
s3, combining a plurality of predetermined spectral features and a plurality of image features in each group of feature sets, and respectively obtaining final features corresponding to each feature set;
s4, respectively carrying out normalization processing on the three final features, inputting the final features after the normalization processing into a trained classifier, and respectively obtaining three corresponding classification results;
the trained classifier is obtained by training the classifier through the final feature for training;
the final features for training include: the method comprises the steps that spectrum data used for training and extracted from image data used for training, corresponding to each type of wood in multiple types of wood, and three groups of final features which are extracted from the image data used for training and are combined with feature sets used for training and corresponding to the type of wood;
and S5, determining the final recognition result of the wood based on the three classification results.
2. The method according to claim 1, wherein the S3 specifically includes:
combining a plurality of predetermined spectral features and a plurality of image features in each group of feature sets by adopting a formula (1), and respectively obtaining final features corresponding to each feature set, wherein the formula (1) is as follows:
f1 is the final feature corresponding to the feature set;
[yn]a set of a plurality of predetermined spectral features corresponding to the feature set;
[wm]a set of a plurality of predetermined image features corresponding to the feature set;
[yn，wm]a plurality of predetermined spectral features and a plurality of predetermined image features corresponding to the feature set;
wherein A is1Is a first predetermined coefficient;
A2a second predetermined coefficient;
B1a third predetermined coefficient;
B2is a fourth predetermined coefficient.
3. The method according to claim 2, wherein the S2 includes:
s21, preprocessing the spectral data to obtain preprocessed spectral data;
the preprocessing operation sequentially comprises one or more of smoothing operation, first-order differential operation or second-order differential operation, standard variable transformation operation and centralization operation;
and S22, extracting three groups of feature sets aiming at the image data and the spectrum data after the preprocessing operation.
4. The method of claim 3,
the wavelength range in the spectral data comprises visible light and near infrared spectral bands;
the image data is an image scanned in equal proportion to the wood surface to be identified corresponding to the spectral data during collection; the resolution of the image data is greater than 512 x 512 pixels;
the wood surface to be identified is previously ground by sand paper with 400, 800 and 1000 meshes in sequence, and the wood surface is polished by a fiber wheel after being ground.
5. The method of claim 4, further comprising, prior to S1:
a1, acquiring a training set of spectral data and image data;
the training set of the spectral data and the image data comprises spectral data and image data, wherein the spectral data and the image data are used for training and correspond to each type of wood in multiple types of wood;
a2, determining each feature set in three groups of feature sets corresponding to each type of wood in the multiple types of wood according to the training set of the spectrum data and the image data;
a3, respectively acquiring three final features for training corresponding to each type of wood based on each feature set in the three sets of feature sets corresponding to each type of wood;
a4, normalizing the three final features for training corresponding to each type of wood in the multiple types of wood to obtain the final features for training after the normalization, inputting the final features for training after the normalization into a classifier, and training by the classifier until the accuracy of the classifier meets a preset value to finish the training to obtain the trained classifier.
6. The method of claim 5, wherein the A2 comprises:
a21, preprocessing each spectral data for training in the set of spectral data for training in the training set of the spectral data and the image data to obtain preprocessed spectral data for training;
a22, aiming at image data which are used for training and correspond to any type of wood in multiple types of wood, adopting a gray level co-occurrence matrix method to extract Z image features which are used for training and correspond to the type of wood;
a23, respectively acquiring the distribution range of each image feature in the image data for training corresponding to each type of wood by adopting a formula (2) according to the image data for training corresponding to each type of wood in the multiple types of wood and the multiple image features;
the formula (2):
wherein the content of the first and second substances,
w for the R-th wood of multiple types of woodiImage feature distribution range; w is ai，minIs wiA minimum value in the image feature data;
wi，maxis wiA maximum value in the image feature data;
n is the number of types of wood included in the training set of spectral and image data;
wiis the ith image feature in the plurality of image features;
a24, respectively acquiring the intersection of each image feature in the wood of multiple types by adopting a formula (3) based on the distribution range of each image feature in the image data which is corresponding to each type of wood and used for training;
the formula (3) is:
wNw for N types of woodiIntersection of image feature data;
wi，min Nw for N types of woodiA minimum value in an intersection of image feature data;
wi，max Nw for N types of woodiA maximum value in an intersection of image feature data;
a25, acquiring image feature difference coefficients of the wood materials of various types by adopting a formula (4) based on the intersection of each image feature of the wood materials of various types;
the formula (4) is:
wherein Z is the number of image features; cNA difference coefficient of image characteristics for N types of wood;
and A26, determining each feature in the three sets of feature sets corresponding to each type of wood in the multiple types of wood based on the difference coefficient.
7. The method of claim 6, wherein said A26 comprises:
a261, if the difference coefficient, satisfies 0.8 < CNIf the spectrum characteristic is less than or equal to 1, aiming at the preprocessed spectrum data set for training, respectively determining the spectrum characteristics corresponding to the preset threshold T in the Relief-F algorithm when the threshold T is 0.25, 0.55 and 0.75 by adopting the Relief-F algorithm;
if the difference coefficient satisfies 0 < CNIf the sum is less than or equal to 0.6, respectively determining spectral characteristics corresponding to preset threshold values T in a Relief-F algorithm when the threshold values T are 0.35, 0.65 and 0.85 by adopting the Relief-F algorithm according to the preprocessed spectral data set sum for training;
if the difference coefficient satisfies 0.6 < CNIf the spectrum data set is less than or equal to 0.8, respectively determining a Relief-F algorithm by adopting the Relief-F algorithm according to the preprocessed spectrum data set for trainingRespectively corresponding spectral characteristics when the preset threshold value T is 0.45, 0.8 and 0.95;
a262, calculating a plurality of image features in image data used for training by adopting a Pearson correlation coefficient algorithm, and acquiring a correlation coefficient p among the image features;
a263, if the difference coefficient, satisfy 0.8 < CNIf the absolute value of the correlation coefficient P is less than or equal to 1, respectively determining the image characteristics of which the absolute value of the correlation coefficient P is less than 0.25, the image characteristics of which the absolute value of the correlation coefficient P is less than 0.55 and the image characteristics of which the absolute value of the correlation coefficient P is less than 0.75;
if the difference coefficient satisfies 0 < CNIf the absolute value of the correlation coefficient p is less than or equal to 0.6, respectively determining the image characteristics of which the absolute value of the correlation coefficient p is less than 0.35, the image characteristics of which the absolute value of the correlation coefficient p is less than 0.65 and the image characteristics of which the absolute value of the correlation coefficient p is less than 0.85;
if the difference coefficient satisfies 0.6 < CNIf the absolute value of the correlation coefficient p is less than or equal to 0.8, respectively determining the image characteristics of which the absolute value of the correlation coefficient p is less than 0.45, the image characteristics of which the absolute value of the correlation coefficient p is less than 0.8 and the image characteristics of which the absolute value of the correlation coefficient p is less than 0.95;
a264, if the difference coefficient, satisfies 0.8 < CNDetermining three groups of feature sets corresponding to each type of wood in the multiple types of wood when the number of the wood is less than or equal to 1;
wherein, three sets of feature sets corresponding to each type of wood in the multiple types of wood comprise: a first group of feature sets, a second group of feature sets, and a third group of feature sets;
the first set of feature sets includes: when a preset threshold value T in a Relief-F algorithm is 0.25, corresponding spectral characteristics and image characteristics with the absolute value of the correlation coefficient p less than 0.25 are obtained;
the second set of feature sets includes: when a preset threshold value T in a Relief-F algorithm is 0.55, corresponding spectral characteristics and image characteristics with the absolute value of the correlation coefficient p less than 0.55 are obtained;
the third set of feature sets comprises: when a preset threshold value T in a Relief-F algorithm is 0.75, corresponding spectral characteristics and image characteristics with the absolute value of the correlation coefficient p less than 0.75 are obtained;
if the difference coefficient satisfies 0 < CNDetermining three groups of feature sets corresponding to each type of wood in the multiple types of wood, wherein the three groups of feature sets are less than or equal to 0.6;
wherein, three sets of feature sets corresponding to each type of wood in the multiple types of wood comprise: a fourth group of feature sets, a fifth group of feature sets and a sixth group of feature sets;
the fourth set of features comprises: when a preset threshold value T in a Relief-F algorithm is 0.35, corresponding spectral characteristics and image characteristics with the absolute value of the correlation coefficient p less than 0.35 are obtained;
the fifth set of features comprises: when a preset threshold value T in a Relief-F algorithm is 0.65, corresponding spectral characteristics and image characteristics with the absolute value of the correlation coefficient p less than 0.65 are obtained;
the sixth set of feature sets comprises: when the value of T in the Relief-F algorithm is 0.85, the corresponding spectral characteristic and the image characteristic of which the absolute value of the correlation coefficient p is less than 0.85 are obtained;
if the difference coefficient satisfies 0.6 < CNDetermining three groups of feature sets corresponding to each type of wood in the multiple types of wood, wherein the three groups of feature sets are less than or equal to 0.8;
wherein, three sets of feature sets corresponding to each type of wood in the multiple types of wood comprise: a seventh group of feature sets, an eighth group of feature sets, and a ninth group of feature sets;
the seventh set of features comprises: when a preset threshold value T in a Relief-F algorithm is 0.45, corresponding spectral characteristics and image characteristics with the absolute value of the correlation coefficient p less than 0.45 are obtained;
the eighth set of features comprises: when a preset threshold value T in a Relief-F algorithm is 0.8, corresponding spectral characteristics and image characteristics with the absolute value of the correlation coefficient p less than 0.8 are obtained;
the ninth set of features includes: when a preset threshold value T in a Relief-F algorithm is 0.95, the corresponding spectral characteristics and the image characteristics of which the absolute value of the correlation coefficient p is less than 0.95 are obtained.
8. The method according to claim 7, wherein the a3 specifically comprises:
aiming at each group of feature set corresponding to each type of wood, adopting a formula (5) to obtain corresponding final features for training;
the formula (5) is:
f2 is the final feature for training corresponding to the set of feature sets;
[ys]all spectral feature sets used for training in the set of feature sets;
[wg]all image feature sets used for training in the set of feature sets;
[ys，wg]for all spectral features used for training and image features used for training in the set of feature sets;
9. The method of claim 8,
the three groups of feature sets extracted for the spectral data and the image data in the S2 are full of the difference coefficient0.8 < CNUnder the condition of less than or equal to 1, the following are respectively: a first group of feature sets, a second group of feature sets, and a third group of feature sets;
the three sets of feature sets extracted for the spectral data and the image data in the S2 satisfy 0 < C in the difference coefficientNUnder the condition of less than or equal to 0.6, the following are respectively: a fourth group of feature sets, a fifth group of feature sets and a sixth group of feature sets;
the three groups of feature sets extracted for the spectral data and the image data in the S2 satisfy 0.6 < C in the difference coefficientNUnder the condition of less than or equal to 0.8, the following are respectively: a seventh set of features, an eighth set of features, and a ninth set of features.
10. The method according to claim 9, wherein the S5 specifically includes: and judging whether the three classification results are the same, and if at least two classification results in the three classification results are the same, taking the same at least two classification results as final identification results.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111616170.9A CN114397250B (en) | 2021-12-27 | 2021-12-27 | Wood identification method based on spectral features and image features |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111616170.9A CN114397250B (en) | 2021-12-27 | 2021-12-27 | Wood identification method based on spectral features and image features |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114397250A true CN114397250A (en) | 2022-04-26 |
| CN114397250B CN114397250B (en) | 2024-02-13 |
Family
ID=81227024
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111616170.9A Active CN114397250B (en) | 2021-12-27 | 2021-12-27 | Wood identification method based on spectral features and image features |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114397250B (en) |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101996328A (en) * | 2010-10-14 | 2011-03-30 | 浙江农林大学 | Wood identification method |
| CN106770003A (en) * | 2016-11-21 | 2017-05-31 | 无锡迅杰光远科技有限公司 | Wood Identification Method and system based on near-infrared spectrum technique |
| CN108280440A (en) * | 2018-02-09 | 2018-07-13 | 三亚中科遥感研究所 | A kind of fruit-bearing forest recognition methods and system |
| CN108458978A (en) * | 2018-03-13 | 2018-08-28 | 山东农业大学 | Based on the seeds multispectral remote sensing recognition methods that sensitive band and band combination are optimal |
| CN109870421A (en) * | 2019-03-27 | 2019-06-11 | 东北林业大学 | It is a kind of based on visible light/near-infrared spectrum analysis incrementally timber varieties of trees classifying identification method |
| CN110348277A (en) * | 2018-11-30 | 2019-10-18 | 浙江农林大学 | A kind of tree species image-recognizing method based under natural background |
| CN111898649A (en) * | 2020-07-07 | 2020-11-06 | 西南林业大学 | Wood tree species classification and identification method based on visible light or near infrared spectrum analysis |
| CN112926702A (en) * | 2019-12-06 | 2021-06-08 | 李雯毓 | Active light source type object material identification system and method |
-
2021
- 2021-12-27 CN CN202111616170.9A patent/CN114397250B/en active Active
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101996328A (en) * | 2010-10-14 | 2011-03-30 | 浙江农林大学 | Wood identification method |
| CN106770003A (en) * | 2016-11-21 | 2017-05-31 | 无锡迅杰光远科技有限公司 | Wood Identification Method and system based on near-infrared spectrum technique |
| CN108280440A (en) * | 2018-02-09 | 2018-07-13 | 三亚中科遥感研究所 | A kind of fruit-bearing forest recognition methods and system |
| CN108458978A (en) * | 2018-03-13 | 2018-08-28 | 山东农业大学 | Based on the seeds multispectral remote sensing recognition methods that sensitive band and band combination are optimal |
| CN110348277A (en) * | 2018-11-30 | 2019-10-18 | 浙江农林大学 | A kind of tree species image-recognizing method based under natural background |
| CN109870421A (en) * | 2019-03-27 | 2019-06-11 | 东北林业大学 | It is a kind of based on visible light/near-infrared spectrum analysis incrementally timber varieties of trees classifying identification method |
| CN112926702A (en) * | 2019-12-06 | 2021-06-08 | 李雯毓 | Active light source type object material identification system and method |
| CN111898649A (en) * | 2020-07-07 | 2020-11-06 | 西南林业大学 | Wood tree species classification and identification method based on visible light or near infrared spectrum analysis |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114397250B (en) | 2024-02-13 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108074231B (en) | Magnetic sheet surface defect detection method based on convolutional neural network | |
| CN106126585B (en) | The unmanned plane image search method combined based on quality grading with perceived hash characteristics | |
| CN101059425A (en) | Method and device for identifying different variety green tea based on multiple spectrum image texture analysis | |
| CN109522924A (en) | A kind of broad-leaf forest wood recognition method based on single photo | |
| CN111738367B (en) | Part classification method based on image recognition | |
| CN104990892A (en) | Spectrum image lossless identification model establishing method for seeds and seed identification method | |
| CN110991434B (en) | Self-service terminal certificate identification method and device | |
| CN110702648A (en) | Fluorescent spectrum pollutant classification method based on non-subsampled contourlet transformation | |
| CN110443303B (en) | Image segmentation and classification-based intelligent identification method for coal-rock microcomponents | |
| CN103745242A (en) | Cross-equipment biometric feature recognition method | |
| CN113870202A (en) | Far-end chip defect detection system based on deep learning technology | |
| CN110136113B (en) | Vagina pathology image classification method based on convolutional neural network | |
| CN113792666A (en) | Concrete classification method and system based on scanning electron microscope images | |
| CN114397250B (en) | Wood identification method based on spectral features and image features | |
| Bremananth et al. | Wood species recognition using GLCM and correlation | |
| Chen et al. | Convolutional neural network-based pure paint pigment identification using hyperspectral images | |
| CN114936840A (en) | Intelligent identification method for power business work order information based on image classification and OCR technology | |
| CN113269136B (en) | Off-line signature verification method based on triplet loss | |
| CN112364844B (en) | Data acquisition method and system based on computer vision technology | |
| CN111563536B (en) | Bamboo strip color self-adaptive classification method based on machine learning | |
| CN111382703B (en) | Finger vein recognition method based on secondary screening and score fusion | |
| CN114970596A (en) | Raman spectrum analysis-based fish meal adulteration rapid detection method | |
| CN111898680A (en) | Biological identification method based on material inspection multi-view morphological image and deep learning | |
| CN112633399B (en) | Sparse collaborative joint representation pattern recognition method | |
| CN111723223B (en) | Multi-label image retrieval method based on subject inference |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |