CN114882297B - Waste metal extrusion sorting data classification storage method - Google Patents
Waste metal extrusion sorting data classification storage method Download PDFInfo
- Publication number
- CN114882297B CN114882297B CN202210784684.3A CN202210784684A CN114882297B CN 114882297 B CN114882297 B CN 114882297B CN 202210784684 A CN202210784684 A CN 202210784684A CN 114882297 B CN114882297 B CN 114882297B
- Authority
- CN
- China
- Prior art keywords
- feature
- category
- waste metal
- individual
- image
- 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
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/764—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using classification, e.g. of video objects
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/50—Information retrieval; Database structures therefor; File system structures therefor of still image data
- G06F16/55—Clustering; Classification
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T9/00—Image coding
- G06T9/005—Statistical coding, e.g. Huffman, run length coding
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/56—Extraction of image or video features relating to colour
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/74—Image or video pattern matching; Proximity measures in feature spaces
- G06V10/761—Proximity, similarity or dissimilarity measures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
- G06V10/774—Generating sets of training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/82—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks
-
- 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
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Physics & Mathematics (AREA)
- Multimedia (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Databases & Information Systems (AREA)
- Evolutionary Computation (AREA)
- Computing Systems (AREA)
- Software Systems (AREA)
- Medical Informatics (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Data Mining & Analysis (AREA)
- General Engineering & Computer Science (AREA)
- Image Analysis (AREA)
Abstract
The invention relates to a sorting and storing method for waste metal extrusion sorting data, which belongs to the technical field of big data storage and comprises the following steps: collecting waste metal sample images, classifying the waste metal sample images according to categories, and extracting characteristic values of a plurality of characteristics in each waste metal sample image; calculating the importance of each feature in each category according to the feature variability and the feature conflict of each feature of the waste metal sample image in each category; determining the fuzziness of each feature in each category according to the dispersion degree and the importance degree of each feature of the waste metal sample image in each category; and taking the probability that each possible value of each feature of the waste metal sample image in each category appears in all feature values of the feature in the category as Huffman coding, and carrying out coding compression on each feature of the waste metal sample image in each category. The invention can improve the data compression rate and the transmission efficiency and can save the storage space under the condition of ensuring the feature accuracy.
Description
Technical Field
The invention belongs to the technical field of big data storage, and particularly relates to a sorting and storing method for waste metal extrusion sorting data.
Background
The waste metal refers to metal fragments and scraps discarded by the metallurgical industry and the metal processing industry, scrapped metal objects for equipment renewal and the like, and also includes metal objects such as metal packaging containers and waste vehicles recovered from municipal waste. The waste metal is a resource, and the recycling of the waste metal resource can promote sustainable development and relieve the resource and environmental pressure caused by ore mining.
In order to recycle the waste metal, the extruded waste metal material needs to be sorted and recycled according to the type, and the sorted waste metal sample image data needs to be sorted and stored. Data storage has strong practical significance, and data management, retrieval and the like can be facilitated only by adopting a reasonable data storage mode. Data storage from a macroscopic perspective, roughly encompasses three broad categories of technologies: B-TREE, LSM & SSTable, columnar storage. In the prior art, data of waste metal sample images are classified and stored through Huffman coding, and due to the fact that the types of the stored data are more, a Huffman tree is large, the coding length is long, the storage space is wasted, and the decompression efficiency is low.
Disclosure of Invention
The invention provides a classification storage method for waste metal extrusion sorting data, and aims to solve the problems that the data of waste metal sample images are classified and stored through Huffman coding at present, and due to the fact that the types of the stored data are more, the Huffman tree is large, the coding length is long, the storage space is wasted, and the decompression efficiency is low.
The invention discloses a sorting and storing method for waste metal extrusion sorting data, which adopts the following technical scheme: the method comprises the following steps:
collecting waste metal sample images, classifying the waste metal sample images according to categories, and extracting characteristic values of a plurality of characteristics in each waste metal sample image; wherein a plurality of features of each image of the scrap metal specimen are the same;
determining the feature variability of each feature of the waste metal sample images in each class according to the difference between each feature of the waste metal sample images in each class and the feature in the rest classes;
calculating the feature correlation between each feature of the waste metal sample image in each category and all other features except the feature in the category, and calculating the feature conflict of each feature of the waste metal sample image in each category by using the obtained all feature correlations;
calculating the importance degree of each feature of the waste metal sample images in each category according to the feature variability and the feature conflict of each feature of the waste metal sample images in each category;
calculating the dispersion degree of each feature of the waste metal sample images in each category according to all feature values of each feature of the waste metal sample images in each category; determining the fuzziness of each feature of the waste metal sample images in each category according to the dispersion degree and the importance degree of each feature of the waste metal sample images in each category;
dividing all characteristic values of each characteristic of the waste metal sample images in each category into a plurality of intervals with the length being ambiguity after being arranged according to the numerical value, selecting the characteristic value positioned in the middle of each interval as all possible values of the characteristic in the category, and counting the probability of each possible value of each characteristic of the waste metal sample images in each category appearing in all the characteristic values of the characteristic in the category;
and obtaining a Huffman coding table according to the probability that each possible value of each feature of the waste metal sample image in each category appears in all feature values of the feature in the category, carrying out coding compression on each feature of the waste metal sample image in each category according to the Huffman coding table, and carrying out classified storage on compressed data according to the category.
Further, the determining the feature variability of each feature of the waste metal sample images in each class according to the difference between each feature of the waste metal sample images in each class and the feature in the rest classes comprises:
acquiring all characteristic values of each characteristic of the waste metal sample image in each category, and calculating a first probability of each characteristic value of each characteristic of the waste metal sample image in each category appearing in all characteristic values of the characteristic in the category; calculating a second probability that each feature value of each feature of the waste metal sample image in each category appears in the other categories except the category;
and determining the feature variability of each feature of the waste metal sample images in each class according to all the obtained first probabilities, all the obtained second probabilities and the number of all feature values of each feature in all the classes.
Further, the calculation formula of the feature variability is shown as follows:
wherein the content of the first and second substances,is shown asImage of scrap metal specimen in individual categoryThe first characteristicCharacteristic value is inIn a category ofA first probability of occurrence in all feature values of the individual feature;is shown asImage of scrap metal specimen in individual categoryA first feature ofThe characteristic value is divided byA second probability that the one category appears outside the remaining categories;second to show images of scrap metal specimensThe number of all feature values of each feature in all categories;is shown asImage of scrap metal specimen in CategoryFeature variability of features.
Further, the calculating the feature correlation between each feature of the image of the waste metal sample in each category and all the features except the feature in the category comprises:
acquiring all characteristic values of each characteristic of each waste metal sample image in each category;
calculating the characteristic value mean value and the characteristic value standard deviation of each characteristic of all the waste metal sample images in each category;
and calculating the feature correlation of each feature of the waste metal sample image in each category and all the features except the feature in the category according to the obtained all feature values, the feature value mean value and the feature value standard deviation.
Further, the calculation formula of the feature correlation is shown as follows:
wherein the content of the first and second substances,is shown asIn a category ofFirst of the images of the sheet scrap metal specimenA feature value of the individual feature;is shown asAll waste metal sample images in the individual categoryMean value of eigenvalues of individual features;is shown asAll waste metal sample images in the individual categoryA standard deviation of eigenvalues of the individual features;is shown asThe total number of images of the waste metal samples contained in each category;is shown asIn a category ofDrawing of waste metal specimenThe first of an imageA feature value of the individual feature;is shown asAll waste metal sample images in each categoryMean value of eigenvalues of individual features;is shown asAll waste metal sample images in the individual categoryA standard deviation of eigenvalues of the individual features;is shown asImage of scrap metal specimen in individual categoryA feature andthe correlation of individual features.
Further, the calculation formula of the feature conflict is shown as follows:
wherein the content of the first and second substances,is shown asImage of scrap metal specimen in individual classA feature anda correlation of the individual features;representing the total number of features in each extracted waste metal sample image;is shown asImage of scrap metal specimen in individual categoryFeature conflicts of individual features.
Further, the calculation formula of the importance is shown as follows:
wherein the content of the first and second substances,is shown asImage of scrap metal specimen in CategoryFeature variability of features;is shown asImage of scrap metal specimen in individual categoryFeature conflicts of individual features;is shown asImage of scrap metal specimen in individual categoryThe importance of the individual features.
Further, the calculation formula of the dispersion degree is shown as the following formula:
wherein the content of the first and second substances,denotes the firstImage of scrap metal specimen in individual categoryA first feature ofCharacteristic value is inIn a category ofA first probability of occurrence in all feature values of the individual feature;is shown asImage of scrap metal specimen in individual categoryThe first characteristicA characteristic value;is shown asImage of scrap metal specimen in individual categoryA first feature ofA characteristic value;is shown asImage of scrap metal specimen in individual categoryThe total eigenvalue number of the individual features;is shown asImage of scrap metal specimen in individual categoryThe dispersion of individual features.
Further, the calculation formula of the ambiguity is shown as follows:
wherein the content of the first and second substances,is shown asImage of scrap metal specimen in individual categoryA degree of dispersion of the features;is shown asImage of scrap metal specimen in individual categoryThe importance of the individual features;is shown asImage of scrap metal specimen in individual categoryThe ambiguity of the individual features.
Further, the extracting the feature values of the plurality of features in each waste metal sample image comprises:
and extracting the characteristic values of all color characteristics and all texture characteristics in each waste metal sample image.
The invention has the beneficial effects that:
the invention provides a sorting and storing method for waste metal extrusion sorting data. The method comprises the steps of firstly sorting waste metal sample images through a training neural network, extracting the characteristics of the waste metal sample images, obtaining the importance of the characteristics in corresponding categories according to the characteristic variability and the characteristic conflict in different categories, distributing the fuzziness to each characteristic in each category according to the importance and the dispersity, and obtaining a Huffman coding table by combining the fuzziness to realize the classified storage of the waste metal extrusion sorting data.
The invention can improve the data compression rate and the transmission efficiency. By Huffman coding, shorter bits are distributed to the features with larger occurrence frequency, and longer bits are distributed to the features with smaller occurrence frequency, so that the data compression rate is improved, and the transmission efficiency is improved.
The invention saves the storage space under the condition of ensuring the feature accuracy. And distributing an ambiguity parameter to each feature according to the importance and the dispersity, and reducing possible kinds of values of the features to be coded through the ambiguity parameter under the condition of ensuring the accuracy of the features so as to save storage space.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to these drawings without creative efforts.
Fig. 1 is a flow chart illustrating the general steps of an embodiment of the sorting and storing method for scrap metal by extrusion.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
An embodiment of the sorting and storing method for the waste metal extrusion sorting data of the invention is shown in fig. 1, and the method comprises the following steps:
s1, collecting waste metal sample images, classifying the waste metal sample images according to categories, and extracting characteristic values of a plurality of characteristics in each waste metal sample image; wherein a plurality of features of each image of the scrap metal specimen are identical.
The invention collects the images of the waste metal samples through the electronic equipment arranged on the conveyor belt. The electronic device includes: and a frame. The detection device comprises: the device is arranged above the conveyor belt and comprises an imaging device and a light source, wherein the imaging device is used for imaging the waste metal in the detection area of the conveyor belt and outputting imaging data, and the light source is used for supplementing brightness; the induction device is arranged on the conveyor belt and the rack and used for inducing the waste metal and outputting an induction signal; and the central processing unit is in electric signal connection with the imaging equipment and the sensing device and is used for receiving the sensing signal and realizing corresponding data processing and control output.
The invention trains the neural network to sort the waste metal sample images. The method comprises the steps of artificially labeling the collected waste metal sample images, training a neural network, performing mode matching by adopting a multilayer perceptron neural network algorithm, further realizing automatic sorting of the waste metals, obtaining the sorting result of the waste metal sample images, collecting the waste metal sample images and sorting the waste metal sample images according to the categories. And after the waste metal sample images are classified, counting the number of sample images contained in different classes.
The method extracts the characteristic values of all color characteristics and all texture characteristics in each waste metal sample image. Obtaining R, G, B, H, S, I color component mean values of each waste metal sample image, and recording the mean values as characteristic values of color features; for a gray level image corresponding to R, G, B channels, a gray level LBP co-occurrence matrix is constructed based on an LBP equivalent mode, 11 statistical characteristics of the gray level co-occurrence matrix are obtained, and the statistical characteristics are respectively texture second-order distance, texture contrast, texture correlation, texture entropy, texture variance, co-occurrence and mean value, co-occurrence and variance, inverse difference moment, co-occurrence difference variance, co-occurrence and entropy and co-occurrence difference entropy and are recorded as characteristic values of texture characteristics.
Each waste metal sample image has a corresponding characteristic value of the color characteristic and a corresponding characteristic value of the texture characteristic. For example, if there are 6 statistical features for the color features and 11 statistical features for the texture features, there are 6 feature values for the color features and 11 feature values for the texture features in each waste metal sample image. For the feature of the texture correlation degree of the texture feature, each waste metal sample image has a feature value, so that all feature values of the feature of the texture correlation degree in each waste metal sample image are calculated for later-stage calculation, and all feature values of each feature in each waste metal sample image are calculated in the same way.
And S2, determining the feature variability of each feature of the waste metal sample images in each class according to the difference between each feature of the waste metal sample images in each class and the feature in the rest classes.
Acquiring all characteristic values of each characteristic of the waste metal sample image in each category, and calculating a first probability of each characteristic value of each characteristic of the waste metal sample image in each category appearing in all characteristic values of the characteristic in the category; calculating a second probability that each feature value of each feature of the waste metal sample image in each category appears in the other categories except the category; and determining the feature variability of each feature of the waste metal sample images in each class according to all the obtained first probabilities, all the obtained second probabilities and the number of all feature values of each feature in all the classes.
For any feature of the waste metal sample image in any category, the greater the difference with the feature in the rest categories, the greater the variability of the feature in the category, which indicates that the feature can characterize the category, i.e. contains more information of the category, so that the more important the feature is, the more quality the feature needs to be guaranteed when performing compression storage. The specific steps for calculating the feature variability are as follows: for any feature of the waste metal sample image in any category, counting the distribution condition of each feature value of the feature in the category, and similarly, counting the distribution condition of each feature value of the feature in all other categories; and calculating feature variability according to the distribution situation.
The calculation formula of the feature variability is shown as follows:
wherein the content of the first and second substances,is shown asImage of scrap metal specimen in individual categoryA first feature ofCharacteristic value is inIn a category ofA first probability of occurrence in all feature values of the individual feature;is shown asImage of scrap metal specimen in individual categoryA first feature ofCharacteristic value is divided byA second probability that the one category appears outside the remaining categories;second to show images of scrap metal specimensThe number of all feature values of each feature in all categories;is shown asImage of scrap metal specimen in CategoryFeature variability of features.
And S3, calculating the feature correlation between each feature of the waste metal sample image in each category and all the features except the feature in the category, and calculating the feature conflict of each feature of the waste metal sample image in each category by using all the obtained feature correlations.
Acquiring all characteristic values of each characteristic of each waste metal sample image in each category; calculating the characteristic value mean value and the characteristic value standard deviation of each characteristic of all the waste metal sample images in each category; and calculating the feature correlation of each feature of the waste metal sample image in each category and all the features except the feature in the category according to the obtained all feature values, the feature value mean value and the feature value standard deviation.
For any feature of the waste metal sample image in any category, the less relevant the feature is to other features in the category, the greater the conflict of the feature in the category is, which indicates that the feature can characterize the category more, that is, the feature contains more information of the category, so the more important the feature is, the more quality the feature needs to be guaranteed when performing compression storage.
The calculation formula of the characteristic correlation is shown as follows:
wherein the content of the first and second substances,is shown asIn a category ofImage of scrap metal specimenA feature value of the individual feature;is shown asAll waste metal sample images in the individual categoryMean value of eigenvalues of individual features;is shown asAll waste metal sample images in the individual categoryA standard deviation of eigenvalues of the individual features;is shown asTotal number of images of the waste metal samples contained in each category;is shown asIn a category ofImage of scrap metal specimenA feature value of the individual feature;is shown asAll waste metal sample images in the individual categoryMean value of eigenvalues of individual features;is shown asAll waste metal sample images in the individual categoryA standard deviation of eigenvalues of the individual features;is shown asImage of scrap metal specimen in individual categoryA characteristic ofThe correlation of individual features.
The calculation formula of the feature conflict is shown as the following formula:
wherein the content of the first and second substances,is shown asImage of scrap metal specimen in individual categoryA feature anda correlation of the individual features;representing the total number of features in each extracted waste metal sample image;is shown asImage of scrap metal specimen in individual categoryFeature conflicts of individual features.
And S4, calculating the importance degree of each feature of the waste metal sample images in each class according to the feature variability and the feature conflict of each feature of the waste metal sample images in each class.
For example: the texture contrast characteristic in the image of the scrap metal iron isIn the image of other scrap metals areAnd then the scrap metal iron can be accurately separated from the scrap metal based on the texture contrast characteristics. The texture contrast characteristics of the waste metal iron are greatly different from those of other types of waste metals (except the waste metal iron), namely the texture contrast characteristics are important for sorting the waste metal iron. And the texture correlation characteristic in the image of the scrap metal iron isIn the images of other scrap metals, tooThen, the scrap metal iron cannot be sorted out from the scrap metal based on the texture correlation characteristic; therefore, the texture contrast characteristic is important for sorting the scrap iron, and the texture correlation characteristic is not important for sorting the scrap iron.
2) The less relevant a feature in a category is to other features, the more important the feature is to the category, i.e., the more conflicting a feature is, the more important the feature is to the category. For example, in an image of scrap metal iron, the texture correlation feature can be linearly represented by the texture contrast feature and the texture variance feature, and the texture contrast feature cannot be linearly represented by other features, so the texture contrast feature is important for sorting the scrap metal iron, and the texture correlation feature is not important for sorting the scrap metal iron.
Therefore, the importance of each feature is comprehensively measured by combining feature variability and feature conflict, and the greater the feature variability and the feature conflict, the more the feature can characterize the class, that is, the more information of the class is contained, so that the more important the feature is, the more quality the feature needs to be guaranteed when performing compressed storage.
The formula for calculating the importance is shown below:
wherein the content of the first and second substances,is shown asImage of scrap metal specimen in CategoryFeature variability of features;is shown asImage of scrap metal specimen in individual categoryFeature conflicts of individual features;is shown asImage of scrap metal specimen in individual categoryThe importance of the individual features.
S5, calculating the dispersion degree of each feature of the waste metal sample images in each category according to all feature values of each feature of the waste metal sample images in each category; and determining the fuzziness of each feature of the waste metal sample images in each category according to the dispersion degree and the importance degree of each feature of the waste metal sample images in each category.
The classification compression storage of the feature data of the waste metal sample image is realized through the Huffman coding, but the length and the storage space of the Huffman coding are increased along with the increase of the possible values of the features, so that the compression efficiency is reduced. The possible value of the characteristic is further reduced by recording the characteristic value in a certain interval range as a possible value, the size of the interval is recorded as the ambiguity, the greater the ambiguity is, the less the possible value of the characteristic is, and the greater the compression efficiency is. Although increasing the ambiguity can shorten the coding length by reducing the possible values of the features, thereby improving the compression efficiency, the ambiguity affects the accuracy of the features, and the greater the ambiguity, the smaller the accuracy of the features, therefore, the present invention assigns different ambiguities to the features according to the importance of the features, and the less important the ambiguity of the features, the greater the less important the ambiguity of the features. While considering that the larger the divergence that the features themselves have, i.e. the larger the distance between adjacent features, the larger the ambiguity that can be assigned to that feature.
For example: the possible value of the texture contrast characteristic in the image of the scrap metal iron isAnd the minimum interval of the features is 2, and the maximum interval of the features is 4, so that the accuracy of the features can be ensured by averaging the intervals of the texture contrast features in the waste metal iron category without the interval being less than 2. I.e., the greater the distance between adjacent features, the greater the degree of ambiguity that can be assigned to that featureIf the texture contrast feature is important, the importance is subtracted from the interval average value of the texture contrast feature, and the ambiguity is reduced, that is, the interval value of the feature is reduced.
Firstly, calculating the dispersion degree of each feature of the waste metal sample image in each category according to the total feature value of each feature of the waste metal sample image in each category, wherein the calculation formula of the dispersion degree is shown as the following formula:
wherein the content of the first and second substances,is shown asImage of scrap metal specimen in individual categoryA first feature ofCharacteristic value is inIn a category ofA first probability of occurrence in all feature values of the individual feature;is shown asImage of scrap metal specimen in individual categoryA first feature ofA characteristic value;is shown asImage of scrap metal specimen in individual categoryA first feature ofA characteristic value;is shown asImage of scrap metal specimen in individual categoryThe total eigenvalue number of the individual features;is shown asImage of scrap metal specimen in individual categoryThe dispersion of individual features.
The formula for calculating the ambiguity is shown as follows:
wherein the content of the first and second substances,denotes the firstImage of scrap metal specimen in individual categoryA degree of dispersion of the features;is shown asImage of scrap metal specimen in individual categoryThe importance of the individual features;denotes the firstImage of scrap metal specimen in individual categoryThe ambiguity of the individual features.
S6, dividing all the characteristic values of each characteristic of the waste metal sample images in each category into a plurality of intervals with the length as ambiguity after being arranged according to the numerical value, selecting the characteristic value in the middle of each interval as all the possible values of the characteristic in the category, and counting the probability of each possible value of each characteristic of the waste metal sample images in each category appearing in all the characteristic values of the characteristic in the category.
The method comprises the steps of sorting all characteristic values of each characteristic of waste metal sample images in each category from large to small according to numerical values to obtain a sequence, dividing the sequence into a plurality of intervals with the length being ambiguity, and enabling the number of the characteristic values contained in each interval to be equal.
And selecting the characteristic value positioned in the middle of each interval as all possible values of each characteristic of the waste metal sample image in each category, and counting the probability of each possible value of each characteristic of the waste metal sample image in each category appearing in all characteristic values of the characteristic in the category.
S7, obtaining a Huffman coding table according to the probability that each possible value of each feature of the waste metal sample image in each category appears in all feature values of the feature in the category, carrying out coding compression on each feature of the waste metal sample image in each category according to the Huffman coding table, and carrying out classified storage on compressed data according to the category.
The invention obtains a Huffman coding table of characteristics according to the ambiguity of the characteristics, which comprises the following specific steps:
1. and sorting the waste metal sample images to obtain the categories and classification results of the waste metal sample images, and extracting all characteristic values of a plurality of characteristics of each waste metal sample image.
2. And calculating feature variability and feature conflict for each feature of the waste metal sample images in each class, further obtaining the importance of each feature in each class, and distributing fuzziness to each feature of the waste metal sample images in each class according to the importance and the dispersion degree.
3. And arranging all characteristic values of each characteristic of the waste metal sample images in each category according to the size to obtain a sequence, dividing the sequence into a plurality of intervals with the length being the ambiguity, wherein the number of the characteristic values contained in each interval is equal. And respectively taking the median value of each interval as each possible value of each feature of the waste metal sample image in each class, and counting the probability that each possible value of each feature of the waste metal sample image in each class appears in all feature values of each feature of the waste metal sample image in each class.
4. And constructing a Huffman tree according to the counted probability that each possible value of each feature of the waste metal sample image in each category appears in all feature values of the feature in the category, and further obtaining a Huffman coding table.
Therefore, the ambiguity is distributed to each feature according to the importance degree and the dispersion degree of the feature, and the Huffman coding table is obtained by combining the ambiguity. And (3) coding and compressing each characteristic of the waste metal sample image of each category according to a Huffman coding table, and classifying and storing the compressed data according to the categories, which belongs to the prior art and is not described herein again.
In summary, the invention provides a classification storage method for waste metal extrusion sorting data, which comprises the steps of firstly sorting waste metal sample images through a training neural network, extracting features of the waste metal sample images, obtaining importance of the features in corresponding categories according to feature variability and feature conflicts in different categories, distributing fuzziness to each feature of the waste metal sample images in each category according to the importance and the dispersity, and obtaining a huffman coding table by combining the fuzziness to realize classification storage of the waste metal extrusion sorting data.
The above description is only for the purpose of illustrating the preferred embodiments of the present invention and is not to be construed as limiting the invention, and any modifications, equivalents, improvements and the like that fall within the spirit and principle of the present invention are intended to be included therein.
Claims (2)
1. A sorting and storing method for waste metal extrusion sorting data is characterized by comprising the following steps:
collecting waste metal sample images, classifying the waste metal sample images according to categories, and extracting characteristic values of a plurality of characteristics in each waste metal sample image; wherein a plurality of features of each image of the scrap metal specimen are the same;
determining feature variability of each feature of the images of the scrap metal specimens in each class from differences between each feature of the images of the scrap metal specimens in each class and the feature in the remaining classes, comprising:
acquiring all characteristic values of each characteristic of the waste metal sample image in each category, and calculating a first probability of each characteristic value of each characteristic of the waste metal sample image in each category appearing in all characteristic values of the characteristic in the category; calculating a second probability that each feature value of each feature of the waste metal sample image in each category appears in the other categories except the category;
determining the feature variability of each feature of the waste metal sample image in each class according to all the obtained first probabilities, all the obtained second probabilities and the number of all feature values of each feature in all the classes;
the calculation formula of the feature variability is shown as follows:
wherein the content of the first and second substances,is shown asImage of scrap metal specimen in individual categoryA first feature ofCharacteristic value is inIn a first categoryA first probability of occurrence in all feature values of the individual feature;is shown asImages of scrap metal samples in individual categoriesTo (1) aA first feature ofCharacteristic value is divided byA second probability that the one category appears outside the remaining categories;second to show images of scrap metal specimensThe number of all feature values of each feature in all categories;denotes the firstImage of scrap metal specimen in CategoryFeature variability of features;
calculating the feature correlation of each feature of the waste metal sample image in each category and all the other features except the feature in the category, wherein the feature correlation comprises the following steps:
acquiring all characteristic values of each characteristic of each waste metal sample image in each category;
calculating the characteristic value mean value and the characteristic value standard deviation of each characteristic of all the waste metal sample images in each category;
calculating the feature correlation of each feature of the waste metal sample image in each category and all other features except the feature in the category according to the obtained all feature values, the feature value mean value and the feature value standard deviation;
the calculation formula of the characteristic correlation is shown as follows:
wherein the content of the first and second substances,is shown asIn a category ofImage of scrap metal specimenA feature value of the individual feature;is shown asAll waste metal sample images in the individual categoryMean value of eigenvalues of individual features;is shown asAll waste metal sample images in the individual categoryA standard deviation of eigenvalues of the individual features;is shown asThe total number of images of the waste metal samples contained in each category;is shown asIn a category ofImage of scrap metal specimenA feature value of the individual feature;is shown asAll waste metal sample images in the individual categoryThe mean value of the eigenvalues of the individual characteristics;denotes the firstAll waste metal sample images in the individual categoryThe characteristic value standard deviation of each characteristic;is shown asImage of scrap metal specimen in individual categoryA feature anda correlation of the individual features;
calculating the feature conflict of each feature of the waste metal sample image in each category by using the obtained all feature correlations;
the calculation formula of the feature conflict is shown as the following formula:
wherein the content of the first and second substances,is shown asImage of scrap metal specimen in individual categoryA feature anda correlation of the individual features;representing the total number of features in each extracted waste metal sample image;is shown asImage of scrap metal specimen in individual classFeature conflicts of individual features;
calculating the importance degree of each feature of the waste metal sample images in each category according to the feature variability and the feature conflict of each feature of the waste metal sample images in each category;
the formula for calculating the importance is shown below:
wherein the content of the first and second substances,is shown asImage of scrap metal specimen in CategoryFeature variability of features;is shown asImage of scrap metal specimen in individual categoryFeature conflicts of individual features;is shown asImage of scrap metal specimen in individual categoryThe importance of the individual features;
calculating the dispersion degree of each feature of the waste metal sample images in each category according to all feature values of each feature of the waste metal sample images in each category;
the formula for calculating the degree of dispersion is shown below:
wherein the content of the first and second substances,is shown asImage of scrap metal specimen in individual categoryA first feature ofCharacteristic value is inIn a category ofA first probability of occurrence in all feature values of the individual feature;is shown asImage of scrap metal specimen in individual categoryA first feature ofA characteristic value;is shown asImage of scrap metal specimen in individual categoryA first feature ofA characteristic value;is shown asImage of scrap metal specimen in individual categoryThe total eigenvalue number of the individual features;denotes the firstImage of scrap metal specimen in individual categoryA degree of dispersion of the features;
determining the fuzziness of each feature of the waste metal sample image in each category according to the dispersion degree and the importance degree of each feature of the waste metal sample image in each category;
the formula for calculating the ambiguity is shown as follows:
wherein the content of the first and second substances,is shown asImage of scrap metal specimen in individual categoryA degree of dispersion of the features;is shown asImage of scrap metal specimen in individual categoryThe importance of the individual features;is shown asImage of scrap metal specimen in individual categoryThe ambiguity of the individual feature;
dividing all characteristic values of each characteristic of the waste metal sample images in each category into a plurality of intervals with the length being ambiguity after being arranged according to the numerical value, selecting the characteristic value positioned in the middle of each interval as all possible values of the characteristic in the category, and counting the probability of each possible value of each characteristic of the waste metal sample images in each category appearing in all the characteristic values of the characteristic in the category;
and obtaining a Huffman coding table according to the probability that each possible value of each feature of the waste metal sample image in each category appears in all feature values of the feature in the category, carrying out coding compression on each feature of the waste metal sample image in each category according to the Huffman coding table, and carrying out classified storage on compressed data according to the category.
2. The sorting and storing method for the waste metal extrusion sorting data as claimed in claim 1, wherein the extracting the feature values of the plurality of features in each waste metal sample image comprises:
and extracting the characteristic values of all color characteristics and all texture characteristics in each waste metal sample image.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210784684.3A CN114882297B (en) | 2022-07-06 | 2022-07-06 | Waste metal extrusion sorting data classification storage method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210784684.3A CN114882297B (en) | 2022-07-06 | 2022-07-06 | Waste metal extrusion sorting data classification storage method |
Publications (2)
Publication Number | Publication Date |
---|---|
CN114882297A CN114882297A (en) | 2022-08-09 |
CN114882297B true CN114882297B (en) | 2022-09-20 |
Family
ID=82683086
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202210784684.3A Active CN114882297B (en) | 2022-07-06 | 2022-07-06 | Waste metal extrusion sorting data classification storage method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN114882297B (en) |
Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113761262A (en) * | 2021-09-03 | 2021-12-07 | 奇安信科技集团股份有限公司 | Image search type determining method, system and image search method |
Family Cites Families (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
BR9702155A (en) * | 1996-03-15 | 1999-07-20 | Philips Electronics Nv | Process of encoding information signal blocks and apparatus for forming encoded information signal blocks |
NL2017769B1 (en) * | 2016-11-11 | 2018-05-24 | Scrapscanner B V | Process and apparatus for scrap metal scanning |
CN108114909B (en) * | 2016-11-29 | 2019-10-11 | 中国科学院沈阳自动化研究所 | A kind of old metal intelligence storting apparatus and method based on Libs system |
US10898928B2 (en) * | 2018-03-27 | 2021-01-26 | Huron Valley Steel Corporation | Vision and analog sensing scrap sorting system and method |
-
2022
- 2022-07-06 CN CN202210784684.3A patent/CN114882297B/en active Active
Patent Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113761262A (en) * | 2021-09-03 | 2021-12-07 | 奇安信科技集团股份有限公司 | Image search type determining method, system and image search method |
Also Published As
Publication number | Publication date |
---|---|
CN114882297A (en) | 2022-08-09 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN111462488B (en) | Intersection safety risk assessment method based on deep convolutional neural network and intersection behavior characteristic model | |
US20190228527A1 (en) | System and method of classification of biological particles | |
CN101814147B (en) | Method for realizing classification of scene images | |
CN114445387A (en) | Fiberboard quality classification method based on machine vision | |
CN108717867A (en) | Disease forecasting method for establishing model and device based on Gradient Iteration tree | |
CN114550018A (en) | Nutrition management method and system based on deep learning food image recognition model | |
CN114511718B (en) | Intelligent management method and system for materials for building construction | |
KR20220156603A (en) | Method and electronic device for automated waste management | |
CN112580647A (en) | Stacked object oriented identification method and system | |
CN114694178A (en) | Method and system for monitoring safety helmet in power operation based on fast-RCNN algorithm | |
CN109347719B (en) | Image spam filtering method based on machine learning | |
CN115019111A (en) | Data processing method for Internet literary composition creation works | |
CN117565284A (en) | Automatic control system and method for PVC film processing | |
CN114882297B (en) | Waste metal extrusion sorting data classification storage method | |
CN111242170A (en) | Food inspection and detection item prediction method and device | |
CN111144021A (en) | Fuel cell service life prediction method and system | |
KR20110062274A (en) | Apparatus and method for selecting optimal database by using the maximal concept strength recognition techniques | |
Busin et al. | Color space selection for color image segmentation by spectral clustering | |
CN111126419B (en) | Dot clustering method and device | |
CN115082449A (en) | Electronic component defect detection method | |
Fahrurozi et al. | Texture Features and Statistical Features for Wood Types Classification System | |
CN116992267B (en) | Regional population gender identification method and system based on signaling data | |
CN116630725B (en) | Multi-dimensional screening-based garbage classification processing method, device, equipment and medium | |
CN114580569A (en) | Construction waste material visual identification method based on feature code fusion | |
CN117853937B (en) | Rice disease identification method and system based on secondary color cluster analysis |
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 |