CN112767387B - Automatic wood image recognition method based on block gradient weighting - Google Patents
Automatic wood image recognition method based on block gradient weighting Download PDFInfo
- Publication number
- CN112767387B CN112767387B CN202110134275.4A CN202110134275A CN112767387B CN 112767387 B CN112767387 B CN 112767387B CN 202110134275 A CN202110134275 A CN 202110134275A CN 112767387 B CN112767387 B CN 112767387B
- Authority
- CN
- China
- Prior art keywords
- image
- wood
- cross
- wood cross
- recognition
- 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.)
- Active
Links
- 239000002023 wood Substances 0.000 title claims abstract description 125
- 238000000034 method Methods 0.000 title claims abstract description 60
- 238000007781 pre-processing Methods 0.000 claims abstract description 12
- 238000012549 training Methods 0.000 claims abstract description 11
- 230000000903 blocking effect Effects 0.000 claims abstract description 8
- 238000013527 convolutional neural network Methods 0.000 claims abstract description 8
- 238000000638 solvent extraction Methods 0.000 claims abstract description 6
- 230000011218 segmentation Effects 0.000 claims abstract description 5
- 238000012937 correction Methods 0.000 claims description 8
- 238000001914 filtration Methods 0.000 claims description 6
- 238000012545 processing Methods 0.000 claims description 5
- 239000006002 Pepper Substances 0.000 claims description 2
- 238000013135 deep learning Methods 0.000 description 3
- 238000012360 testing method Methods 0.000 description 3
- 238000010586 diagram Methods 0.000 description 2
- 230000007613 environmental effect Effects 0.000 description 2
- 238000011156 evaluation Methods 0.000 description 2
- 238000000605 extraction Methods 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 238000012216 screening Methods 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 239000011521 glass Substances 0.000 description 1
- 230000005484 gravity Effects 0.000 description 1
- 238000005286 illumination Methods 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 230000035764 nutrition Effects 0.000 description 1
- 235000016709 nutrition Nutrition 0.000 description 1
- 238000012567 pattern recognition method Methods 0.000 description 1
- 150000003839 salts Chemical class 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0004—Industrial image inspection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2415—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on parametric or probabilistic models, e.g. based on likelihood ratio or false acceptance rate versus a false rejection rate
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T3/00—Geometric image transformations in the plane of the image
- G06T3/40—Scaling of whole images or parts thereof, e.g. expanding or contracting
- G06T3/4007—Scaling of whole images or parts thereof, e.g. expanding or contracting based on interpolation, e.g. bilinear interpolation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/80—Geometric correction
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/11—Region-based segmentation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/90—Determination of colour characteristics
-
- 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
- G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10024—Color image
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20021—Dividing image into blocks, subimages or windows
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30108—Industrial image inspection
- G06T2207/30161—Wood; Lumber
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Data Mining & Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Computing Systems (AREA)
- Biomedical Technology (AREA)
- General Health & Medical Sciences (AREA)
- Computational Linguistics (AREA)
- Biophysics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Molecular Biology (AREA)
- Health & Medical Sciences (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Probability & Statistics with Applications (AREA)
- Evolutionary Biology (AREA)
- Multimedia (AREA)
- Quality & Reliability (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses an automatic wood image recognition method based on block gradient weighting, which comprises the steps of preprocessing a wood cross section image; partitioning the wood cross-sectional image; training the sub-images after the segmentation through a convolutional neural network model; and adopting different gradient values from the edge to the center of the wood cross-section image as weights of different sub-region image classification scores, increasing the proportion of the center region in the whole wood cross-section image classification score, and converting the weighted score into a final probability value. According to the technical scheme provided by the invention, through blocking the image, finer image characteristics can be extracted, the recognition accuracy of wood is improved, meanwhile, the robustness of the model is enhanced by adopting a gradient weighting method, and the fault tolerance and generalization capability of the model are improved.
Description
Technical Field
The invention relates to the field of image recognition, in particular to an automatic wood image recognition method based on block gradient weighting.
Background
At present, the wood species identification is mainly carried out by professionals under laboratory environment, and the macroscopic and microscopic characteristics of the wood are observed by means of instruments and equipment such as a magnifying glass, a microscope and the like, and then the wood species is determined after the comparison with a standard sample. Each wood forms a unique macroscopic and microscopic structure in the growth process, but the wood has a plurality of types, and even the same wood has a certain difference in macroscopic and microscopic structure due to the conditions of places, climate, nutrition and the like. The phenomena of similarity and intra-species difference increase the difficulty of wood species identification, and in addition, the expert with wood taxonomy knowledge in China's wood trade and supervision line is rare, wood or wood products are frequently transacted, and the artificial-based wood species identification method has the problems of strong specialization, heavy task, long period, high risk, non-real-time property and the like, and cannot meet the requirements of real-time property and high efficiency of wood supervision, so that a rapid and accurate wood species identification method is needed.
With rapid development and application of image processing technology, wood species identification by using images has become a research hotspot in recent years. The wood species recognition method based on the image can be summarized into two types, one type is to use a traditional mode recognition method, and the other type is to use a deep learning method. The method comprises the steps of preprocessing a wood image by using a traditional mode recognition method, extracting and screening effective recognition features, and carrying out classification recognition by using a machine learning method. The image characteristics in the traditional pattern recognition method are obtained through a large number of experiments and experience screening, and a higher recognition rate can be generally obtained in a test set, but the robustness and generalization capability of the model are poor. In recent years, the deep learning method is successfully applied to image recognition, and has the advantages of autonomous learning, feature diversity, high accuracy, strong model robustness and generalization capability and the like. The image-based deep learning identification method mainly utilizes a convolutional neural network to perform autonomous distributed feature extraction on wood images, performs more abstract and generalized expression on features by deepening the network layer number, and then classifies the features by utilizing the extracted high-order feature information. However, most of existing wood image recognition researches aim at wood images shot under a laboratory microscope, the types of recognized woods are few, the woods with similar cross sections are lack, the robustness and generalization capability of the model are poor, and the rapid and accurate recognition cannot be achieved.
Therefore, in order to solve the above-mentioned problems, there is a need for an automatic wood image recognition method based on block gradient weighting.
Disclosure of Invention
In view of the above, it is necessary to provide an automatic wood image recognition method based on block gradient weighting, so as to accurately and rapidly automatically recognize wood images, and ensure relatively high recognition rate, and the technical scheme provided by the invention is as follows:
The invention provides a timber image automatic identification method based on block gradient weighting, which comprises the following steps:
S1, preprocessing a wood cross-section image;
s2, partitioning the wood cross-section image;
s3, training the segmented sub-images through a convolutional neural network model;
s4, adopting different gradient values from the edge to the center of the wood cross-section image as weights of sub-image classification scores of different areas, increasing the proportion of the center area in the whole wood cross-section image classification score, and converting the weighted score into a final probability value; wherein,
The step of preprocessing the wood cross-sectional image in S1 comprises:
S11, performing color correction on the wood cross-section image;
s12, carrying out data enhancement on the corrected wood cross-section image;
the step of processing the wood cross-sectional image by the segmentation in S2 comprises:
s21, dividing the pretreated wood cross-section image to obtain a plurality of sub-images;
S22, unifying size pixels of each sub-image through a bilinear interpolation method.
Further, step S11 includes performing color correction on the wood cross-sectional image by a gray world method, calculating R, G, B three-channel gain coefficients and gray average values from the entire wood cross-sectional image and the pixel expectation values of R, G, B three channels, and performing respective R, G, B three-channel component adjustment on each pixel in the wood cross-sectional image according to the gain coefficients and the gray average values.
Further, step S12 includes data enhancement of the corrected wood cross-sectional images using horizontal flipping, vertical flipping, and salt-and-pepper noise addition to bring each wood image training sample size into a specified number range.
Further, the partitioning process employs a method of 5×5 and/or 7×7 and/or 10×10.
Further, the convolutional neural network model is ResNet model 101.
Further, the weighted score is converted to a final probability value by Softmax method.
Further, the preprocessing also comprises Gaussian filtering and/or Laplacian filtering and/or histogram equalization.
The invention has the following advantages:
a. The image is processed in a blocking way, so that finer characteristics can be extracted, the recognition and classification of the cross-section images of the wood are facilitated, and the recognition accuracy of the wood is improved;
b. Color correction is adopted to eliminate the color cast influence caused by the image acquisition equipment and the environmental difference, so that high-quality images are provided for subsequent model training, and the wood image recognition rate is more accurate;
c. the block gradient weighting method enhances the robustness of the model and improves the fault tolerance and generalization capability of the model.
Drawings
Fig. 1 is a network structure diagram of a method for automatically identifying a wood image according to an embodiment of the present invention;
fig. 2 is a line diagram of wood recognition accuracy of different partitioning methods in the automatic wood image recognition method according to the embodiment of the present invention.
Detailed Description
In order that those skilled in the art will better understand the present invention, a technical solution in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in which it is apparent that the described embodiments are only some embodiments of the present invention, not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the present invention without making any inventive effort, shall fall within the scope of the present invention.
It should be noted that the terms "first," "second," and the like in the description and the claims of the present invention and the above figures are used for distinguishing between similar objects and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used may be interchanged where appropriate such that the embodiments of the invention described herein may be implemented in sequences other than those illustrated or otherwise described herein. Furthermore, the terms "comprise" and "have," as well as any variations thereof, are intended to cover a non-exclusive inclusion.
In one embodiment of the present invention, there is provided an automatic wood image recognition method based on block gradient weighting, as shown in fig. 1, the automatic wood image recognition method including the steps of:
S1, preprocessing a wood cross-section image;
s2, partitioning the wood cross-section image;
S3, training an image through a convolutional neural network model;
S4, adopting different gradient values from the edge to the center of the wood cross-section image as weights of different sub-region image classification scores, increasing the proportion of the center region in the whole wood cross-section image classification score, and converting the weighted score into a final probability value;
wherein the step of preprocessing the wood cross-sectional image in S1 comprises:
S11, performing color correction on the wood cross-section image;
s12, carrying out data enhancement on the corrected wood cross-section image;
the step of processing the wood cross-sectional image by the segmentation in S2 comprises:
s21, dividing the pretreated wood cross-section image to obtain a plurality of sub-images;
S22, unifying size pixels of each sub-image through a bilinear interpolation method.
Specifically, in a first step, wood cross-sectional images are pre-treated.
Since the wood cross-sectional image features are greatly affected by external illumination and acquisition equipment, in one embodiment of the invention, the gray world method is used for performing color correction on the wood cross-sectional image, and the specific process is as follows:
Firstly, calculating the overall pixel expected value of the wood cross-sectional image and the pixel expected value of R, G, B three channels respectively; secondly, respectively calculating gain coefficients of R, G, B three channels according to the overall pixel expected value of the wood cross-section image and the pixel expected value of R, G, B three channels, and calculating a gray average value of R, G, B three channels; and finally, according to the gain coefficient and the gray average value of R, G, B channels, each pixel in the wood cross-section image is subjected to respective R, G, B three-channel component adjustment so as to eliminate color cast influence caused by image acquisition equipment and environmental differences.
Besides the color correction of the wood cross-section image, the corrected wood cross-section image can be subjected to data enhancement, so that the sample image is expanded, and the robustness and generalization capability of the model can be improved. In one embodiment of the invention, the data enhancement adopts the methods of horizontal overturn, vertical overturn and spiced salt noise addition, so that each wood image training sample size is in a specified quantity range.
It should be noted that the above preprocessing method is only an example, and the preprocessing may also use methods such as contrast broadening, logarithmic transformation, density layering, histogram equalization, gaussian filtering, and laplace filtering, which do not limit the protection scope of the present invention.
And secondly, blocking the wood cross-sectional image.
Because the macro structure of the cross section of the wood has self-similar characteristics, in order to obtain more image characteristics and training sample size, the preprocessed cross section image of the wood needs to be segmented to obtain a plurality of sub-images, and then the size pixels of each sub-image are unified by a bilinear interpolation method, and the specific process is as follows:
In one embodiment of the invention, an image of a wood cross-section is image segmented using three segmentation methods, 5 x 5, 7 x 7 and 10 x 10, to obtain 25, 49 and 100 sub-images, which are then scaled to a uniform size, 224 x 224 pixels, by bilinear interpolation.
It should be noted that the above-mentioned blocking method is only an example, and other blocking methods, such as 6×6, may be used, which does not limit the protection scope of the present invention.
And thirdly, training the image through a convolutional neural network model.
In order to evaluate the effectiveness of the wood recognition model provided by the invention, in one embodiment of the invention, an Accuracy (Accuracy), a Recall (Recall), an Average Accuracy (Average Accuracy) and an Average Recall (AVERAGE RECALL) are selected as evaluation indexes, and the calculation formulas of the evaluation indexes are as follows:
Wherein, accuracy represents the Accuracy of model recognition, TP represents the number of correct recognition of a certain wood, FP represents the number of incorrect recognition as a certain wood, recall represents the Recall rate of model recognition, FN represents the number of incorrect recognition as another kind of wood, average Accuracy represents the Average Accuracy of model recognition, C represents the number of wood kinds, AVERAGE RECALL represents the Average Recall rate of model recognition.
In one embodiment of the present invention, seven models of the same wood cross-sectional image training set VggNet, googleNet, denseNet, mobileNetv3, resNet50, resNet101, and ResNet were used to test 32 types of wood, and the recognition results are shown in table 1.
Table 1 recognition results of different models on 32 wood cross-section artwork
Wherein the average accuracy and recall of the ResNet series of three models ResNet50, resNet101, and ResNet were over 85%, indicating that the ResNet model is superior to the other models in wood feature extraction, and that the average recognition accuracy of ResNet101 is 4.2% and 1.6% higher than ResNet, resNet152, respectively, in the ResNet network model. Therefore ResNet is chosen as a base model for wood species identification.
It should be noted that, in an embodiment of the present invention, aiming at the block division manner in the second step, the ResNet model is trained on the sub-images obtained by performing three blocks of original image, 5×5, 7×7 and 10×10 on the wood cross-section image, so as to obtain recognition accuracy results of 32 wood tests, as shown in fig. 2, based on the ResNet model of the block division, the recognition accuracy and average recognition accuracy of each wood are better than those of the ResNet model based on the original image, which indicates that the block division strategy can effectively improve the wood recognition accuracy; furthermore, the average recognition accuracy obtained based on the ResNet model of 7×7 blocks is better than that of 5×5 blocks and 10×10 blocks, while the recognition accuracy based on the ResNet model of the original image is always kept at about 92%, because more detail information is lost when the scaled original image is sent to the ResNet network, and more image contents can be reserved in sub-images after the blocks. From the point of view of the stability of the classification result in fig. 2, the recognition result of the original cross-section of the wood and the 5×5 blocks fluctuates greatly, which means that the robustness of the model is not good, and therefore, the 7×7 block strategy is finally selected.
And fourthly, adopting different gradient values from the edge to the center of the wood cross-section image as weights of different sub-region image classification scores, increasing the proportion of the center region in the whole wood cross-section image classification score, and converting the weighted score into a final probability value.
For the ResNet model based on the original image, more effective information is lost in the wood cross-section image due to the problems of blurring and deformation of the image edge, and meanwhile, image noise is increased, so that the recognition result is negatively influenced. Since each sub-image in the ResNet model is given the same weight, the contribution to the recognition result of each sub-image is the same, and the influence of edge blurring and deformation on the image recognition result cannot be reduced. In order to further improve the accuracy of wood image recognition, in one embodiment of the present invention, different gradient values (0.6, 0.7, 0.8, 0.9 and 1 in order) are given to the edge to the center of the wood cross-sectional image as weights of the different sub-area image classification scores, and the recognition results are shown in table 2. It should be noted that the specific gravity of the central region in the overall wood cross-sectional image classification score needs to be increased to reduce the influence of edge blurring and deformation phenomena on the recognition result.
TABLE 2 Wood identification results of ResNet model based on block gradient weighting
As can be seen from table 2, the average recognition rate and average recall rate of ResNet model based on the block gradient weighting method are higher than those of ResNet model based on original image and ResNet model based on block, which indicates that the method adopting the block gradient weighting method can effectively improve the accuracy of wood recognition.
In one embodiment of the invention, the original Softmax classification layer of ResNet model is improved, and the specific process is as follows:
First, different gradient values (0.6, 0.7, 0.8, 0.9 and 1 in order) are used as weights for the classification scores of the different sub-area images from the edge to the center of the wood cross-sectional image.
The weighted scores of the classes are then converted to final probability values using the Softmax method, which is shown in the following equation:
Wherein S represents the weighted score value, n represents the number of rows of the block, m represents the number of columns of the block, Z represents the classification score value of each sub-image, Y represents the final probability value, and C represents the total number of categories of wood.
The probability values of various woods are obtained through the method, so that the wood images are automatically identified, and the purpose of accurate classification is achieved.
According to the technical scheme provided by the invention, the method for processing the wood image by blocks is adopted, so that finer and abstract features can be extracted, fine classification of the cross section image of the wood is facilitated, and the recognition accuracy of the wood is improved; meanwhile, a blocking gradient weighting method is adopted, so that the robustness of the model is enhanced, and the fault tolerance and generalization capability of the model are improved.
The foregoing description is only of the preferred embodiments of the present invention and is not intended to limit the scope of the claims, and all equivalent structures or equivalent processes using the descriptions and drawings of the present invention are directly or indirectly applied to other related technical fields, which are also included in the scope of the present invention.
Claims (5)
1. The automatic wood image recognition method based on the block gradient weighting is characterized by comprising the following steps of:
S1, preprocessing a wood cross-section image;
s2, partitioning the wood cross-section image;
s3, training the segmented sub-images through a convolutional neural network model;
s4, adopting different gradient values from the edge to the center of the wood cross-section image as weights of sub-image classification scores of different areas, increasing the proportion of the center area in the whole wood cross-section image classification score, and converting the weighted score into a final probability value; wherein,
The step of preprocessing the wood cross-sectional image in S1 comprises:
s11, performing color correction on the wood cross-sectional image, wherein the color correction is performed on the wood cross-sectional image through a gray world method, a R, G, B three-channel gain coefficient and a gray average value are calculated according to the whole wood cross-sectional image and the R, G, B three-channel pixel expected values, and each pixel in the wood cross-sectional image is subjected to adjustment of R, G, B three-channel components according to the gain coefficient and the gray average value;
S12, carrying out data enhancement on the corrected wood cross-section image, wherein the data enhancement is carried out on the corrected wood cross-section image by adopting a method of horizontal overturning, vertical overturning and adding salt-and-pepper noise, so that the training sample size of each wood image is in a specified number range;
the step of processing the wood cross-sectional image by the segmentation in S2 comprises:
s21, dividing the pretreated wood cross-section image to obtain a plurality of sub-images;
s22, unifying size pixels of each sub-image by a bilinear interpolation method;
The weighted score is converted to a final probability value by the following formula:
Wherein S represents the weighted score value, n represents the number of rows of the block, m represents the number of columns of the block, Z represents the classification score value of each sub-image, Y represents the final probability value, and C represents the total number of categories of wood.
2. The automatic wood image recognition method according to claim 1, wherein the blocking process employs a5 x 5 and/or 7 x 7 and/or 10 x 10 blocking method.
3. The method for automatically identifying a wood image according to claim 1, wherein the convolutional neural network model is ResNet model 101.
4. The method of automatic recognition of a wood image according to claim 1, wherein the weighted score is converted into a final probability value by a Softmax method.
5. The method for automatic recognition of a wood image according to claim 1, wherein the preprocessing further comprises gaussian filtering and/or laplace filtering and/or histogram equalization.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110134275.4A CN112767387B (en) | 2021-01-29 | 2021-01-29 | Automatic wood image recognition method based on block gradient weighting |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110134275.4A CN112767387B (en) | 2021-01-29 | 2021-01-29 | Automatic wood image recognition method based on block gradient weighting |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112767387A CN112767387A (en) | 2021-05-07 |
| CN112767387B true CN112767387B (en) | 2024-04-30 |
Family
ID=75704420
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110134275.4A Active CN112767387B (en) | 2021-01-29 | 2021-01-29 | Automatic wood image recognition method based on block gradient weighting |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112767387B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116843184B (en) * | 2023-07-27 | 2024-02-20 | 中国公路工程咨询集团有限公司 | Highway construction major risk source identification method and early warning system based on neural network |
Citations (6)
| 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 |
| CN106548165A (en) * | 2016-11-28 | 2017-03-29 | 中通服公众信息产业股份有限公司 | A kind of face identification method of the convolutional neural networks weighted based on image block |
| CN109063713A (en) * | 2018-07-20 | 2018-12-21 | 中国林业科学研究院木材工业研究所 | A kind of timber discrimination method and system based on the study of construction feature picture depth |
| CN110210362A (en) * | 2019-05-27 | 2019-09-06 | 中国科学技术大学 | A kind of method for traffic sign detection based on convolutional neural networks |
| CN110503051A (en) * | 2019-08-27 | 2019-11-26 | 西南林业大学 | A kind of precious timber identifying system and method based on image recognition technology |
| CN111208138A (en) * | 2020-02-28 | 2020-05-29 | 天目爱视（北京）科技有限公司 | Intelligent wood recognition device |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP5417368B2 (en) * | 2011-03-25 | 2014-02-12 | 株式会社東芝 | Image identification apparatus and image identification method |
| US9430829B2 (en) * | 2014-01-30 | 2016-08-30 | Case Western Reserve University | Automatic detection of mitosis using handcrafted and convolutional neural network features |
-
2021
- 2021-01-29 CN CN202110134275.4A patent/CN112767387B/en active Active
Patent Citations (6)
| 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 |
| CN106548165A (en) * | 2016-11-28 | 2017-03-29 | 中通服公众信息产业股份有限公司 | A kind of face identification method of the convolutional neural networks weighted based on image block |
| CN109063713A (en) * | 2018-07-20 | 2018-12-21 | 中国林业科学研究院木材工业研究所 | A kind of timber discrimination method and system based on the study of construction feature picture depth |
| CN110210362A (en) * | 2019-05-27 | 2019-09-06 | 中国科学技术大学 | A kind of method for traffic sign detection based on convolutional neural networks |
| CN110503051A (en) * | 2019-08-27 | 2019-11-26 | 西南林业大学 | A kind of precious timber identifying system and method based on image recognition technology |
| CN111208138A (en) * | 2020-02-28 | 2020-05-29 | 天目爱视（北京）科技有限公司 | Intelligent wood recognition device |
Non-Patent Citations (1)
| Title |
|---|
| 基于梯度融合规则的医学图像融合方法;汪亮亮;张贵仓;贾雯晓;;计算机应用研究;20170614(第07期);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112767387A (en) | 2021-05-07 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108416266B (en) | Method for rapidly identifying video behaviors by extracting moving object through optical flow | |
| CN108510504B (en) | Image segmentation method and device | |
| CN107808132A (en) | A kind of scene image classification method for merging topic model | |
| CN107516103B (en) | Image classification method and system | |
| CN107038416B (en) | Pedestrian detection method based on binary image improved HOG characteristics | |
| CN111222519B (en) | Construction method, method and device of hierarchical colored drawing manuscript line extraction model | |
| CN109190625B (en) | Large-angle perspective deformation container number identification method | |
| CN106340016A (en) | DNA quantitative analysis method based on cell microscope image | |
| CN111598001A (en) | Apple tree pest and disease identification method based on image processing | |
| CN113537182A (en) | Automatic identification method and system for metaphase mitosis microscopic image of chromosome | |
| CN112767387B (en) | Automatic wood image recognition method based on block gradient weighting | |
| CN115908371A (en) | Plant leaf disease and insect pest degree detection method based on optimized segmentation | |
| CN116071339A (en) | Product defect identification method based on improved whale algorithm optimization SVM | |
| CN115731257A (en) | Leaf form information extraction method based on image | |
| CN116071560A (en) | Fruit identification method based on convolutional neural network | |
| CN111563542A (en) | Automatic plant classification method based on convolutional neural network | |
| CN113610024B (en) | Multi-strategy deep learning remote sensing image small target detection method | |
| CN114612399A (en) | Picture identification system and method for mobile phone appearance mark | |
| CN110874835B (en) | Crop leaf disease resistance identification method and system, electronic equipment and storage medium | |
| CN111126185B (en) | Deep learning vehicle target recognition method for road gate scene | |
| CN112257730A (en) | Plant pest image identification method, device, equipment and storage medium | |
| CN113516193B (en) | Image processing-based red date defect identification and classification method and device | |
| CN114862883A (en) | Target edge extraction method, image segmentation method and system | |
| CN108197663A (en) | Based on the calligraphy work image classification method to pairing set Multi-label learning | |
| CN110570437B (en) | Electric power channel automatic inspection data processing method based on boundary identification |
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 |