CN117911796B - Intelligent data processing system and method based on image recognition - Google Patents
Intelligent data processing system and method based on image recognition Download PDFInfo
- Publication number
- CN117911796B CN117911796B CN202410303596.6A CN202410303596A CN117911796B CN 117911796 B CN117911796 B CN 117911796B CN 202410303596 A CN202410303596 A CN 202410303596A CN 117911796 B CN117911796 B CN 117911796B
- Authority
- CN
- China
- Prior art keywords
- image data
- data
- calculating
- collected
- quality
- 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
- 238000012545 processing Methods 0.000 title claims abstract description 25
- 238000000034 method Methods 0.000 title claims abstract description 20
- 238000004364 calculation method Methods 0.000 claims abstract description 63
- 238000012958 reprocessing Methods 0.000 claims abstract description 5
- 230000007613 environmental effect Effects 0.000 claims abstract description 4
- 230000002159 abnormal effect Effects 0.000 claims description 60
- 238000007781 pre-processing Methods 0.000 claims description 12
- 239000011159 matrix material Substances 0.000 claims description 11
- 238000003672 processing method Methods 0.000 claims description 11
- 230000000007 visual effect Effects 0.000 claims description 9
- 238000013480 data collection Methods 0.000 claims description 6
- 238000005516 engineering process Methods 0.000 claims description 5
- 230000015572 biosynthetic process Effects 0.000 claims description 3
- 238000001914 filtration Methods 0.000 claims description 3
- 238000013136 deep learning model Methods 0.000 description 3
- 238000013527 convolutional neural network Methods 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 230000018109 developmental process Effects 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 238000005286 illumination Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000003708 edge detection Methods 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 230000011218 segmentation Effects 0.000 description 1
- 238000012549 training Methods 0.000 description 1
Landscapes
- Image Processing (AREA)
Abstract
The invention discloses an intelligent data processing system and method based on image recognition, and belongs to the technical field of data processing. The invention uses the vision sensor to collect the environmental data to generate the image data, and preprocesses the collected image data; then extracting features, and classifying the image data according to the extracted features; after classifying the image data, calculating the balance degree and visibility of the collected image data, comparing and analyzing the image data collected in real time with the classified image data, and calculating the accuracy of classifying the image data; calculating data quality according to the calculated balance, definition and accuracy of a group of data, and performing early warning when the data quality is lower than a threshold value; and after the data quality early warning occurs, carrying out inversion according to a data quality calculation principle, positioning factors influencing the data quality early warning, outputting positioning data, and reprocessing.
Description
Technical Field
The invention relates to the technical field of data processing, in particular to an intelligent data processing system and method based on image recognition.
Background
With the development of hardware and the progress of algorithms, the field of computer vision has made tremendous progress. From the initial simple edge detection and shape recognition, computer vision has become the core in modern technology to the point that complex tasks such as object detection, image classification, and semantic segmentation can be achieved today. In computer vision, deep learning is used as a machine learning method based on an artificial neural network, and revolutionary progress is brought to image recognition. Particularly, the development of Convolutional Neural Networks (CNNs) greatly improves the accuracy of image recognition and classification. The deep learning model can automatically learn features and modes from the image by learning a large amount of marking data, so that more accurate and intelligent image recognition is realized, and the intelligent data processing of the image recognition requires a large amount of marking data to train the deep learning model. With the popularity of the internet and various sensors, a large amount of image data is continuously accumulated and marked. These data become important resources for training deep learning models, so that the models have a significant improvement in recognition accuracy and generalization capability.
Therefore, in the application of computer vision, the strict requirements on the collected data are extremely important, common image data processing modes include acquisition, storage, transmission, analysis, display and the like, most of the image data processing modes cannot verify processing results after preliminary processing of the data, and elements affecting the data quality cannot be accurately searched after the data quality is calculated, and the specific elements are reprocessed; according to the problems, the invention provides an intelligent data processing system and method based on image recognition.
Disclosure of Invention
The invention aims to provide an intelligent data processing system and method based on image recognition, which are used for solving the problems in the background technology.
In order to solve the technical problems, the invention provides the following technical scheme:
an intelligent data processing method based on image recognition, the method comprising the steps of:
S100, collecting environment data by using a vision sensor to generate image data, and preprocessing the collected image data;
further, the specific steps of preprocessing the collected data are as follows:
S101, detecting the environment in a visual range by using a visual sensor, generating image data of the environment, and setting the number of the collected image data as ，/>Expressed as calculated 1 st, 2 nd b pieces of image data, b is a positive integer;
S102, denoising the collected image data by using a Gaussian filtering technology, identifying pixels in the image data, and filling the pixels at the edge of the image data to the same size.
The influence of interference factors such as illumination, noise and the like on the calculation of the subsequent image data can be reduced by preprocessing the collected image data, and errors are reduced;
S200, extracting features of the preprocessed image data, and classifying the image data according to the extracted features;
Further, the specific steps of classifying the image data according to the extracted features are as follows:
s201, collecting environment image data in history, and setting the type of information contained in each image data as ，/>N pieces of information representing 1,2, and 3..n pieces of information contained in each image data, n being a positive integer; calculating the occurrence frequency of each piece of information in each piece of image data, and selecting the information with the largest occurrence frequency as the characteristic information of the corresponding piece of image data;
S202, according to the mode, after the information type number of all collected historical image data is calculated, j pieces of characteristic information are obtained, and the characteristic set of the image data is constructed by utilizing all the characteristic information as ，The characteristic information is expressed as calculated 1,2 and 3. J, j is a positive integer; dividing the image into j categories according to the characteristic information;
s203, calculating the number of each piece of information contained in each piece of image data collected in real time by the vision sensor, extracting the information with the largest number as corresponding representative information, and finally calculating to obtain the representative information of each piece of image data as ，/>B represents information represented by the calculated 1 st, 2 nd and 3 rd, b is a positive integer;
S204, comparing the representative information and the characteristic information of each image obtained through calculation in sequence, classifying the image data into the category corresponding to the characteristic information when judging that the types of the representative information and the characteristic information are the same, and finally classifying all the collected image data.
The environment is monitored in real time through the vision sensor, the collected data form a group of image data, and the image data are classified, so that the follow-up processes of recognizing and calculating the images in the environment and the like can be more convenient and rapid.
S300, after classifying the image data, calculating the balance degree of the collected image data;
further, the specific steps of calculating the balance degree of the collected image data are as follows:
S301, after classifying each collected image data, respectively calculating the number of the image data in each classification, and taking the calculated number of the image data in each classification as the number of samples; the number of the image data in each classification is calculated as ，/>The number of image data in the 1, 2, 3..j data categories, j being a positive integer;
S302, calculating the average value and standard deviation of the number of the image data of each category, wherein the formulas are respectively as follows:
In the above-mentioned formula(s), For the average value of the number of image data in each category,/>Obtaining standard deviation of the number of the image data of each category for calculation; i has a value of 1 to j;
S303, calculating the difference between the number of the image data in each category and the average value as g_c, comparing the calculated difference between the number of the image data in each category and the average value with the standard deviation, and when g_c > When g_c is less than or equal to/>, judging the image of the corresponding classification as unbalanced image dataWhen the image is classified, the image is classified into balance image data, and the balance image data is classified into x types and y types;
S304, calculating the balance degree of the collected image data according to the number of the unbalanced image data and the balanced image data, wherein the formula is as follows:
In the above-mentioned formula(s), Representing the degree of balance of the collected image data.
When the number of different types of image data is large, the error is possibly too large when the image data is calculated, identified and analyzed, so that the balance degree of the image data is calculated, and the calculated error can be effectively reduced by adjusting the balance degree.
S400, analyzing the collected image data, and calculating the definition of the collected image data;
further, the specific steps of calculating the visibility of the environmental image data are:
S401, extracting pixel values in the collected image data, and setting a pixel formation matrix in each image data as In matrix/>N is the number of rows of the pixel points, M is the number of columns of the pixel points, and the total number of the pixel points in one image data is N multiplied by M;
s402, calculating an average value p_s of pixel values in a matrix, then calculating a difference value of each pixel value and the average value, taking the absolute value of the difference value as p_c, and calculating the definition of each image data, wherein the formula is as follows:
In the above-mentioned formula(s), Representing the sharpness of each image data, b being the total number of collected image data;
S403, collecting image data of information successfully extracted from the history, calculating the definition of the image data in the history, then calculating the average value of the image data in the history, and taking the calculated average value of the definition of the history image data as a threshold value de_y for judging the definition in real time;
s404, calculating definition of the real-time collected image data When/>, compared with the sharpness threshold de_yWhen not less than de_y, judging that the image data is clear, and when/>When < de_y, judging that the image data is not clear; finally, judging to obtain z clear image data in the image data collected in real time;
S405, calculating the visibility of the image data collected in real time, wherein the formula is as follows:
In the above-mentioned formula(s), B is the total number of collected image data for the overall visibility of the image data collected in real time. The definition of each image is calculated, clear images in the collected image data are screened out, the visibility of the collected image data is obtained according to the proportion of the clear images in all the images, the visibility is used as a calculation element of data quality, the effectiveness of the collected image data can be ensured, and the image data which cannot extract information because the visibility is not high is discharged.
S500, comparing and analyzing the image data collected in real time with the classified image data, and calculating the accuracy of classifying the image data;
further, the specific steps of calculating the accuracy of classifying the image data are as follows:
S501, verifying each image data after classification, extracting a pixel point of characteristic information in each image data classification and a pixel point of representing information in the image data, setting the pixel point of the characteristic information to be positioned as (a 1, a 2), calculating a pixel value as K_t, positioning the pixel point of representing information in each image data in the classification as (b 1, b 2), and setting the corresponding pixel value as K;
S502, comparing the calculated pixel point position and corresponding pixel value of the image data representing information with the pixel point position and corresponding pixel value of the characteristic information respectively;
s503, calculating the accuracy of classifying the image data, wherein the formula is as follows:
In the above-mentioned formula(s), Representing the accuracy of image data classification, b represents the total number of image data collected in real time, h1 represents the number of images with the same pixel point position after the comparison of S502, and h2 represents the number of images with the same pixel value after the comparison.
S600, calculating data quality according to the calculated balance degree, definition and accuracy of the whole image data, and performing early warning when the data quality is lower than a threshold value;
further, the specific steps of calculating the data quality are as follows:
S601, calculating the data quality of image data collected in real time, wherein the formula is as follows:
In the above-mentioned formula(s), Representing the data quality of the collected image data,/>For the degree of balance of image data,/>For the visibility of image data,/>Accuracy of classification for image data;
S602, collecting image data which is successfully collected and extracted by a visual sensor in the history and works normally, calculating the data quality of the image data in the history during normal work according to a formula of S601, and calculating an average value to obtain a data quality threshold value Da_y;
s603, judging the image data collected in real time according to the data quality threshold, when When Da_y is not less than the preset value, judging that the quality of the collected and processed image data is qualified, and when Da_y is not less than the preset value, judging that the quality of the collected and processed image data is not more than the preset valueAnd when Da_y is smaller than Da_y, judging that the quality of the collected and processed image data is unqualified, and sending out early warning.
And S700, after the data quality early warning occurs, inverting according to a data quality calculation principle, positioning factors influencing the data quality early warning, outputting positioning data, and reprocessing.
Further, inversion is carried out according to a data quality calculation principle, and the specific steps for locating factors influencing data quality early warning are as follows:
S701, after data quality early warning is carried out on image data collected in real time, the calculated balance degree, visibility and accuracy are used as elements during data quality calculation, sorting and marking are carried out according to the sequence during data quality calculation, and the calculated elements of the data quality are sorted and marked to obtain balance degree 1-visibility 2-accuracy 3;
s702, judging an abnormal element by using an inequality:
In the above inequality, f is a value corresponding to each element for calculating the data quality, e is a reference number of the first element input into the above inequality, and T is a reference number of the element judged to be abnormal by inputting into the above formula;
S702, accumulating the calculated elements according to the sorting input by utilizing the inequality, stopping calculating when the inequality is satisfied, judging the last input element as an abnormal element Y1, and extracting the label of the last input abnormal element Y1; if the element is not completely input into the inequality, the calculated value in the previous inequality is clear, the input of the element after the abnormal element Y1 is continued to be calculated until all the elements are input into the inequality, and the calculation is stopped;
s703, finally calculating to obtain abnormal elements as ，R is a positive integer for the 1 st, 2 nd, 3 rd abnormal elements calculated; and each abnormal element is correspondingly marked as/>，/>A reference numeral representing the 1 st, 2 nd, 3 rd abnormal elements;
s704, outputting the labels of all the abnormal elements obtained by calculation after sending out data quality early warning.
According to the method, the abnormal elements with the data quality lower than the threshold value are examined, all the abnormal elements can be searched through sequential calculation, and then the abnormal elements are reprocessed and extracted, so that the quality of the image data can be accurately and rapidly improved, the effectiveness of the image data is improved, and the subsequent work is more perfect and rapid.
An intelligent data processing system based on image recognition comprises a data collection module, a data processing module, an element calculation module, a data quality calculation module, an abnormal element positioning module and an output module;
The data collection module is used for collecting image data of the environment by using the vision sensor and preprocessing the image data;
The data processing module is used for extracting characteristics of the collected and preprocessed data, and classifying the collected image data according to the characteristics;
The element calculation module is used for calculating the values of various indexes of the image data collected in real time;
The data quality calculation module is used for calculating the data quality of the image data by taking the calculated values of various indexes of the image data as data elements;
the abnormal element positioning module is used for reversely deducting and judging abnormal elements which lead to sending the early warning according to the multiples of the element and the data quality threshold value when sending the data abnormal early warning, and positioning the abnormal elements;
The output module is used for outputting the label of the abnormal element obtained by calculation after sending out early warning.
The element calculation module comprises a balance calculation unit, a visibility calculation unit and an accuracy calculation unit;
The balance degree calculating unit is used for calculating the number of the images in each classified image data, judging the difference of the number of the images in each image data and calculating the balance degree of the image data;
the visibility calculation unit is used for calculating pixel values in the image data, judging the definition of each collected image data, and calculating the visibility of the image data according to the ratio of the overall definition of the collected image data;
The accuracy calculating unit is used for calculating the classification of the image data for the second time and judging the accuracy of the first classification.
Compared with the prior art, the invention has the following beneficial effects:
1. the invention can effectively show the validity and the integrity of the collected image data by respectively calculating the balance degree, the visibility and the accuracy of the image data and taking three values as elements for calculating the quality of the image data.
2. The invention calculates the data quality of the collected image data and compares the data quality with the historical data quality threshold value, so that the availability of the image data collected in real time can be judged, unusable image data can be timely checked out for reprocessing, and the working efficiency of the system is improved.
3. According to the invention, inversion is carried out according to the calculation principle of the image data quality, the abnormal element with low image data quality is accurately searched and positioned, the position of the abnormal element when the data quality is calculated is obtained, and finally the positioning information is output, so that the system can adjust the abnormal element more quickly, and the working speed of the system is increased.
Drawings
The accompanying drawings are included to provide a further understanding of the invention and are incorporated in and constitute a part of this specification, illustrate the invention and together with the embodiments of the invention, serve to explain the invention. In the drawings:
FIG. 1 is a block diagram of an intelligent data processing system based on image recognition in accordance with the present invention;
fig. 2 is a schematic diagram of steps of an intelligent data processing method based on image recognition.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
Referring to fig. 1-2, the present invention provides the following technical solutions:
an intelligent data processing method based on image recognition, the method comprising the steps of:
S100, collecting environment data by using a vision sensor to generate image data, and preprocessing the collected image data;
the specific steps of preprocessing the collected data are as follows:
S101, detecting the environment in a visual range by using a visual sensor, generating image data of the environment, and setting the number of the collected image data as ，/>Expressed as calculated 1 st, 2 nd b pieces of image data, b is a positive integer;
S102, denoising the collected image data by using a Gaussian filtering technology, identifying pixels in the image data, and filling the pixels at the edge of the image data to the same size.
The influence of interference factors such as illumination, noise and the like on the calculation of the subsequent image data can be reduced by preprocessing the collected image data, and errors are reduced;
S200, extracting features of the preprocessed image data, and classifying the image data according to the extracted features;
the specific steps of classifying the image data according to the extracted features are as follows:
s201, collecting environment image data in history, and setting the type of information contained in each image data as ，/>N pieces of information representing 1,2, and 3..n pieces of information contained in each image data, n being a positive integer; calculating the occurrence frequency of each piece of information in each piece of image data, and selecting the information with the largest occurrence frequency as the characteristic information of the corresponding piece of image data;
S202, according to the mode, after the information type number of all collected historical image data is calculated, j pieces of characteristic information are obtained, and the characteristic set of the image data is constructed by utilizing all the characteristic information as ，The characteristic information is expressed as calculated 1,2 and 3. J, j is a positive integer; dividing the image into j categories according to the characteristic information;
s203, calculating the number of each piece of information contained in each piece of image data collected in real time by the vision sensor, extracting the information with the largest number as corresponding representative information, and finally calculating to obtain the representative information of each piece of image data as ，/>B represents information represented by the calculated 1 st, 2 nd and 3 rd, b is a positive integer;
S204, comparing the representative information and the characteristic information of each image obtained through calculation in sequence, classifying the image data into the category corresponding to the characteristic information when judging that the types of the representative information and the characteristic information are the same, and finally classifying all the collected image data.
The environment is monitored in real time through the vision sensor, the collected data form a group of image data, and the image data are classified, so that the follow-up processes of recognizing and calculating the images in the environment and the like can be more convenient and rapid.
S300, after classifying the image data, calculating the balance degree of the collected image data;
The method comprises the following specific steps of calculating the balance degree of the collected image data:
S301, after classifying each collected image data, respectively calculating the number of the image data in each classification, and taking the calculated number of the image data in each classification as the number of samples; the number of the image data in each classification is calculated as ，/>The number of image data in the 1, 2, 3..j data categories, j being a positive integer;
S302, calculating the average value and standard deviation of the number of the image data of each category, wherein the formulas are respectively as follows:
In the above-mentioned formula(s), For the average value of the number of image data in each category,/>Obtaining standard deviation of the number of the image data of each category for calculation; i has a value of 1 to j;
S303, calculating the difference between the number of the image data in each category and the average value as g_c, comparing the calculated difference between the number of the image data in each category and the average value with the standard deviation, and when g_c > When g_c is less than or equal to/>, judging the image of the corresponding classification as unbalanced image dataWhen the image is classified, the image is classified into balance image data, and the balance image data is classified into x types and y types;
S304, calculating the balance degree of the collected image data according to the number of the unbalanced image data and the balanced image data, wherein the formula is as follows:
In the above-mentioned formula(s), Representing the degree of balance of the collected image data.
When the number of different types of image data is large, the error is possibly too large when the image data is calculated, identified and analyzed, so that the balance degree of the image data is calculated, and the calculated error can be effectively reduced by adjusting the balance degree.
S400, analyzing the collected image data, and calculating the definition of the collected image data;
the specific steps of calculating the visibility of the environmental image data are:
S401, extracting pixel values in the collected image data, and setting a pixel formation matrix in each image data as In matrix/>N is the number of rows of the pixel points, M is the number of columns of the pixel points, and the total number of the pixel points in one image data is N multiplied by M;
s402, calculating an average value p_s of pixel values in a matrix, then calculating a difference value of each pixel value and the average value, taking the absolute value of the difference value as p_c, and calculating the definition of each image data, wherein the formula is as follows:
In the above-mentioned formula(s), Representing the sharpness of each image data, b being the total number of collected image data;
S403, collecting image data of information successfully extracted from the history, calculating the definition of the image data in the history, then calculating the average value of the image data in the history, and taking the calculated average value of the definition of the history image data as a threshold value de_y for judging the definition in real time;
s404, calculating definition of the real-time collected image data When/>, compared with the sharpness threshold de_yWhen not less than de_y, judging that the image data is clear, and when/>When < de_y, judging that the image data is not clear; finally, judging to obtain z clear image data in the image data collected in real time;
S405, calculating the visibility of the image data collected in real time, wherein the formula is as follows:
In the above-mentioned formula(s), B is the total number of collected image data for the overall visibility of the image data collected in real time. The definition of each image is calculated, clear images in the collected image data are screened out, the visibility of the collected image data is obtained according to the proportion of the clear images in all the images, the visibility is used as a calculation element of data quality, the effectiveness of the collected image data can be ensured, and the image data which cannot extract information because the visibility is not high is discharged.
S500, comparing and analyzing the image data collected in real time with the classified image data, and calculating the accuracy of classifying the image data;
the specific steps for calculating the accuracy of classifying the image data are as follows:
S501, verifying each image data after classification, extracting a pixel point of characteristic information in each image data classification and a pixel point of representing information in the image data, setting the pixel point of the characteristic information to be positioned as (a 1, a 2), calculating a pixel value as K_t, positioning the pixel point of representing information in each image data in the classification as (b 1, b 2), and setting the corresponding pixel value as K;
S502, comparing the calculated pixel point position and corresponding pixel value of the image data representing information with the pixel point position and corresponding pixel value of the characteristic information respectively;
s503, calculating the accuracy of classifying the image data, wherein the formula is as follows:
In the above-mentioned formula(s), Representing the accuracy of image data classification, b represents the total number of image data collected in real time, h1 represents the number of images with the same pixel point position after the comparison of S502, and h2 represents the number of images with the same pixel value after the comparison.
S600, calculating data quality according to the calculated balance degree, definition and accuracy of the whole image data, and performing early warning when the data quality is lower than a threshold value;
The specific steps for calculating the data quality are as follows:
S601, calculating the data quality of image data collected in real time, wherein the formula is as follows:
In the above-mentioned formula(s), Representing the data quality of the collected image data,/>For the degree of balance of image data,/>For the visibility of image data,/>Accuracy of classification for image data;
S602, collecting image data which is successfully collected and extracted by a visual sensor in the history and works normally, calculating the data quality of the image data in the history during normal work according to a formula of S601, and calculating an average value to obtain a data quality threshold value Da_y;
s603, judging the image data collected in real time according to the data quality threshold, when When Da_y is not less than the preset value, judging that the quality of the collected and processed image data is qualified, and when Da_y is not less than the preset value, judging that the quality of the collected and processed image data is not more than the preset valueAnd when Da_y is smaller than Da_y, judging that the quality of the collected and processed image data is unqualified, and sending out early warning.
And S700, after the data quality early warning occurs, inverting according to a data quality calculation principle, positioning factors influencing the data quality early warning, outputting positioning data, and reprocessing.
Inversion is carried out according to the data quality calculation principle, and the specific steps for locating factors influencing the data quality early warning are as follows:
S701, after data quality early warning is carried out on image data collected in real time, the calculated balance degree, visibility and accuracy are used as elements during data quality calculation, sorting and marking are carried out according to the sequence during data quality calculation, and the calculated elements of the data quality are sorted and marked to obtain balance degree 1-visibility 2-accuracy 3;
s702, judging an abnormal element by using an inequality:
In the above inequality, f is a value corresponding to each element for calculating the data quality, e is a reference number of the first element input into the above inequality, and T is a reference number of the element judged to be abnormal by inputting into the above formula;
S702, accumulating the calculated elements according to the sorting input by utilizing the inequality, stopping calculating when the inequality is satisfied, judging the last input element as an abnormal element Y1, and extracting the label of the last input abnormal element Y1; if the element is not completely input into the inequality, the calculated value in the previous inequality is clear, the input of the element after the abnormal element Y1 is continued to be calculated until all the elements are input into the inequality, and the calculation is stopped;
s703, finally calculating to obtain abnormal elements as ，R is a positive integer for the 1 st, 2 nd, 3 rd abnormal elements calculated; and each abnormal element is correspondingly marked as/>，/>A reference numeral representing the 1 st, 2 nd, 3 rd abnormal elements;
s704, outputting the labels of all the abnormal elements obtained by calculation after sending out data quality early warning.
According to the method, the abnormal elements with the data quality lower than the threshold value are examined, all the abnormal elements can be searched through sequential calculation, and then the abnormal elements are reprocessed and extracted, so that the quality of the image data can be accurately and rapidly improved, the effectiveness of the image data is improved, and the subsequent work is more perfect and rapid.
An intelligent data processing system based on image recognition comprises a data collection module, a data processing module, an element calculation module, a data quality calculation module, an abnormal element positioning module and an output module;
The data collection module is used for collecting image data of the environment by using the vision sensor and preprocessing the image data;
The data processing module is used for extracting characteristics of the collected and preprocessed data, and classifying the collected image data according to the characteristics;
The element calculation module is used for calculating the values of various indexes of the image data collected in real time;
The data quality calculation module is used for calculating the data quality of the image data by taking the calculated values of various indexes of the image data as data elements;
the abnormal element positioning module is used for reversely deducting and judging abnormal elements which lead to sending the early warning according to the multiples of the element and the data quality threshold value when sending the data abnormal early warning, and positioning the abnormal elements;
The output module is used for outputting the label of the abnormal element obtained by calculation after sending out early warning.
The element calculation module comprises a balance calculation unit, a visibility calculation unit and an accuracy calculation unit;
The balance degree calculating unit is used for calculating the number of the images in each classified image data, judging the difference of the number of the images in each image data and calculating the balance degree of the image data;
the visibility calculation unit is used for calculating pixel values in the image data, judging the definition of each collected image data, and calculating the visibility of the image data according to the ratio of the overall definition of the collected image data;
The accuracy calculating unit is used for calculating the classification of the image data for the second time and judging the accuracy of the first classification.
Example 1
The method comprises the steps of monitoring data of a certain environment by using a vision sensor, generating image data from the collected environment data, preprocessing the collected image data, extracting features, classifying the image data according to the extracted features, and classifying the image data as s1=，s2=/>，s3=/>，s4=
The balance degree of the image data is calculated, the average value is 3 according to a formula, the standard deviation is 0.82, and the balance degree is calculated according to the formula:
calculating to obtain the balance degree of the image data as 50%;
let a matrix of pixel values of an image be The definition of the matrix is calculated, and the formula is:
The definition is calculated to be 24, the definition of 16 images is calculated according to the method, the definition threshold value is calculated to be 20 according to the historical data, 14 image data which are larger than the threshold value are obtained through comparison, and the visibility of the image data is calculated according to the formula:
the visibility of the calculated image data is 87.5%;
The accuracy is calculated by comparing the characteristic information and the representative information pixels in each image data after classification according to the following formula:
The accuracy of the calculated image data is 93.8%;
Calculating the data quality of the image data, wherein the formula is as follows:
The calculated result is 77.1%;
Setting a threshold value for knowing the data quality as 78% according to the historical data, judging that the data quality of the collected real-time image data is lower, and sending out early warning;
The three data are ordered and numbered as balance 1-visibility 2-accuracy 3, abnormal elements are found according to the formula,
Firstly, 50% <78% of a balance input formula is obtained, so that the balance is judged to be an abnormal element, the index of the balance is extracted to be 1, then the value in the formula is emptied, 87.5% >78% of a visibility input formula is obtained, the elements are continuously added without meeting the conditions, 181.3% >156% of an accuracy input formula is obtained without meeting the conditions, all the elements are calculated and are stopped, the operation is stopped, the obtained abnormal element is the balance, and the index 1 of the balance is output.
It is noted that relational terms such as first and second, and the like are used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Moreover, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus.
Finally, it should be noted that: the foregoing description is only a preferred embodiment of the present invention, and the present invention is not limited thereto, but it is to be understood that modifications and equivalents of some of the technical features described in the foregoing embodiments may be made by those skilled in the art, although the present invention has been described in detail with reference to the foregoing embodiments. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.
Claims (10)
1. An intelligent data processing method based on image recognition is characterized in that: the method comprises the following steps:
S100, collecting environment data by using a vision sensor to generate image data, and preprocessing the collected image data;
S200, extracting features of the preprocessed image data, and classifying the image data according to the extracted features;
s300, after classifying the image data, calculating the balance degree of the collected whole image data;
s400, analyzing the collected image data, and calculating the visibility of the collected image data;
S500, comparing and analyzing the image data collected in real time with the classified image data, and calculating the accuracy of classifying the image data;
S600, calculating data quality according to the balance degree, the visibility and the accuracy of the overall image data obtained through calculation, and performing early warning when the data quality is lower than a threshold value;
And S700, after the data quality early warning occurs, inverting according to a data quality calculation principle, positioning factors influencing the data quality early warning, outputting positioning data, and reprocessing.
2. The intelligent data processing method based on image recognition according to claim 1, wherein: the specific steps of preprocessing the collected data in S100 are as follows:
S101, detecting the environment in a visual range by using a visual sensor, generating image data of the environment, and setting the number of the collected image data as ，/>Expressed as calculated 1 st, 2 nd b pieces of image data, b is a positive integer;
S102, denoising the collected image data by using a Gaussian filtering technology, identifying pixels in the image data, and filling the pixels at the edge of the image data to the same size.
3. The intelligent data processing method based on image recognition according to claim 1, wherein: the specific step of classifying the image data according to the extracted features in S200 is as follows:
s201, collecting environment image data in history, and setting the type of information contained in each image data as ，/>N pieces of information representing 1,2, and 3..n pieces of information contained in each image data, n being a positive integer; calculating the occurrence frequency of each piece of information in each piece of image data, and selecting the information with the largest occurrence frequency as the characteristic information of the corresponding piece of image data;
S202, according to the mode, after the information type number of all collected historical image data is calculated, j pieces of characteristic information are obtained, and the characteristic set of the image data is constructed by utilizing all the characteristic information as ，/>The characteristic information is expressed as calculated 1,2 and 3. J, j is a positive integer; dividing the image into j categories according to the characteristic information;
s203, calculating the number of each piece of information contained in each piece of image data collected in real time by the vision sensor, extracting the information with the largest number as corresponding representative information, and finally calculating to obtain the representative information of each piece of image data as ，/>B represents information represented by the calculated 1 st, 2 nd and 3 rd, b is a positive integer;
S204, comparing the representative information and the characteristic information of each image obtained through calculation in sequence, classifying the image data into the category corresponding to the characteristic information when judging that the types of the representative information and the characteristic information are the same, and finally classifying all the collected image data.
4. The intelligent data processing method based on image recognition according to claim 1, wherein: the specific steps of calculating the balance degree of the collected image data in S300 are as follows:
S301, after classifying each collected image data, respectively calculating the number of the image data in each classification, and taking the calculated number of the image data in each classification as the number of samples; the number of the image data in each classification is calculated as ，/>The number of image data in the 1, 2, 3..j data categories, j being a positive integer;
S302, calculating the average value and standard deviation of the number of the image data of each category, wherein the formulas are respectively as follows:
In the above-mentioned formula(s), For the average value of the number of image data in each category,/>Obtaining standard deviation of the number of the image data of each category for calculation; i has a value of 1 to j;
S303, calculating the difference between the number of the image data in each category and the average value as g_c, comparing the calculated difference between the number of the image data in each category and the average value with the standard deviation, and when g_c > When g_c is less than or equal to/>, judging the image of the corresponding classification as unbalanced image dataWhen the image is classified, the image is classified into balance image data, and the balance image data is classified into x types and y types;
S304, calculating the balance degree of the collected image data according to the number of the unbalanced image data and the balanced image data, wherein the formula is as follows:
In the above-mentioned formula(s), Representing the degree of balance of the collected image data.
5. The intelligent data processing method based on image recognition according to claim 1, wherein: the specific steps of calculating the visibility of the environmental image data in S400 are:
S401, extracting pixel values in the collected image data, and setting a pixel formation matrix in each image data as In matrix/>N is the number of rows of the pixel points, M is the number of columns of the pixel points, and the total number of the pixel points in one image data is N multiplied by M;
s402, calculating an average value p_s of pixel values in a matrix, then calculating a difference value of each pixel value and the average value, taking the absolute value of the difference value as p_c, and calculating the definition of each image data, wherein the formula is as follows:
In the above-mentioned formula(s), Representing the sharpness of each image data, b being the total number of collected image data;
S403, collecting image data of information successfully extracted from the history, calculating the definition of the image data in the history, then calculating the average value of the image data in the history, and taking the calculated average value of the definition of the history image data as a threshold value de_y for judging the definition in real time;
s404, calculating definition of the real-time collected image data When compared with the definition threshold de_yWhen not less than de_y, judging that the image data is clear, and when/>When < de_y, judging that the image data is not clear; finally, judging to obtain z clear image data in the image data collected in real time;
S405, calculating the visibility of the image data collected in real time, wherein the formula is as follows:
In the above-mentioned formula(s), B is the total number of collected image data for the overall visibility of the image data collected in real time.
6. The intelligent data processing method based on image recognition according to claim 1, wherein: the specific steps of calculating the accuracy of classifying the image data in S500 are as follows:
S501, verifying each image data after classification, extracting a pixel point of characteristic information in each image data classification and a pixel point of representing information in the image data, setting the pixel point of the characteristic information to be positioned as (a 1, a 2), calculating a pixel value as K_t, positioning the pixel point of representing information in each image data in the classification as (b 1, b 2), and setting the corresponding pixel value as K;
S502, comparing the calculated pixel point position and corresponding pixel value of the image data representing information with the pixel point position and corresponding pixel value of the characteristic information respectively;
s503, calculating the accuracy of classifying the image data, wherein the formula is as follows:
In the above-mentioned formula(s), Representing the accuracy of image data classification, b represents the total number of image data collected in real time, h1 represents the number of images with the same pixel point position after the comparison of S502, and h2 represents the number of images with the same pixel value after the comparison.
7. The intelligent data processing method based on image recognition according to claim 1, wherein: the specific steps of calculating the data quality in S600 are as follows:
S601, calculating the data quality of image data collected in real time, wherein the formula is as follows:
In the above-mentioned formula(s), Representing the data quality of the collected image data,/>For the degree of balance of image data,/>For the visibility of image data,/>Accuracy of classification for image data;
S602, collecting image data which is successfully collected and extracted by a visual sensor in the history and works normally, calculating the data quality of the image data in the history during normal work according to a formula of S601, and calculating an average value to obtain a data quality threshold value Da_y;
s603, judging the image data collected in real time according to the data quality threshold, when When Da_y is not less than the preset value, judging that the quality of the collected and processed image data is qualified, and when Da_y is not less than the preset value, judging that the quality of the collected and processed image data is not more than the preset valueAnd when Da_y is smaller than Da_y, judging that the quality of the collected and processed image data is unqualified, and sending out early warning.
8. The intelligent data processing method based on image recognition according to claim 7, wherein: in S700, inversion is performed according to a data quality calculation principle, and the specific steps of locating factors affecting data quality early warning are as follows:
S701, after data quality early warning is carried out on image data collected in real time, the calculated balance degree, visibility and accuracy are used as elements during data quality calculation, sorting and marking are carried out according to the sequence during data quality calculation, and the calculated elements of the data quality are sorted and marked to obtain balance degree 1-visibility 2-accuracy 3;
s702, judging an abnormal element by using an inequality:
In the above inequality, f is a value corresponding to each element for calculating the data quality, e is a reference number of the first element input into the above inequality, and T is a reference number of the element judged to be abnormal by inputting into the above formula;
S702, accumulating the calculated elements according to the sorting input by utilizing the inequality, stopping calculating when the inequality is satisfied, judging the last input element as an abnormal element Y1, and extracting the label of the last input abnormal element Y1; if the element is not completely input into the inequality, the calculated value in the previous inequality is clear, the input of the element after the abnormal element Y1 is continued to be calculated until all the elements are input into the inequality, and the calculation is stopped;
s703, finally calculating to obtain abnormal elements as ，/>R is a positive integer for the 1 st, 2 nd, 3 rd abnormal elements calculated; and each abnormal element is correspondingly marked as，/>A reference numeral representing the 1 st, 2 nd, 3 rd abnormal elements;
s704, outputting the labels of all the abnormal elements obtained by calculation after sending out data quality early warning.
9. An intelligent data processing system based on image recognition is characterized in that: the intelligent data processing system comprises a data collection module, a data processing module, an element calculation module, a data quality calculation module, an abnormal element positioning module and an output module;
The data collection module is used for collecting image data of the environment by using the vision sensor and preprocessing the image data;
The data processing module is used for extracting characteristics of the collected and preprocessed data, and classifying the collected image data according to the characteristics;
The element calculation module is used for calculating the values of various indexes of the image data collected in real time;
The data quality calculation module is used for calculating the data quality of the image data by taking the calculated values of various indexes of the image data as data elements;
the abnormal element positioning module is used for reversely deducting and judging abnormal elements which lead to sending the early warning according to the multiples of the element and the data quality threshold value when sending the data abnormal early warning, and positioning the abnormal elements;
The output module is used for outputting the label of the abnormal element obtained by calculation after sending out early warning.
10. An intelligent data processing system based on image recognition according to claim 9, wherein: the element calculation module comprises a balance calculation unit, a visibility calculation unit and an accuracy calculation unit;
The balance degree calculating unit is used for calculating the number of the images in each classified image data, judging the difference of the number of the images in each image data and calculating the balance degree of the image data;
the visibility calculation unit is used for calculating pixel values in the image data, judging the definition of each collected image data, and calculating the visibility of the image data according to the ratio of the overall definition of the collected image data;
The accuracy calculating unit is used for calculating the classification of the image data for the second time and judging the accuracy of the first classification.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410303596.6A CN117911796B (en) | 2024-03-18 | 2024-03-18 | Intelligent data processing system and method based on image recognition |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410303596.6A CN117911796B (en) | 2024-03-18 | 2024-03-18 | Intelligent data processing system and method based on image recognition |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN117911796A CN117911796A (en) | 2024-04-19 |
| CN117911796B true CN117911796B (en) | 2024-06-21 |
Family
ID=90695063
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202410303596.6A Active CN117911796B (en) | 2024-03-18 | 2024-03-18 | Intelligent data processing system and method based on image recognition |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117911796B (en) |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107610101A (en) * | 2017-08-24 | 2018-01-19 | 昆明理工大学 | A kind of method for measuring digital picture visual balance quality |
| CN117496532A (en) * | 2023-11-09 | 2024-02-02 | 中通服和信科技有限公司 | Intelligent recognition tool based on 0CR |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110163300B (en) * | 2019-05-31 | 2021-04-23 | 北京金山云网络技术有限公司 | Image classification method and device, electronic equipment and storage medium |
-
2024
- 2024-03-18 CN CN202410303596.6A patent/CN117911796B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107610101A (en) * | 2017-08-24 | 2018-01-19 | 昆明理工大学 | A kind of method for measuring digital picture visual balance quality |
| CN117496532A (en) * | 2023-11-09 | 2024-02-02 | 中通服和信科技有限公司 | Intelligent recognition tool based on 0CR |
Also Published As
| Publication number | Publication date |
|---|---|
| CN117911796A (en) | 2024-04-19 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110992349A (en) | Underground pipeline abnormity automatic positioning and identification method based on deep learning | |
| CN108711148B (en) | Tire defect intelligent detection method based on deep learning | |
| CN114663436A (en) | Cross-scale defect detection method based on deep learning | |
| CN111242899B (en) | Image-based flaw detection method and computer-readable storage medium | |
| CN113469951B (en) | Hub defect detection method based on cascade region convolutional neural network | |
| CN116188475B (en) | Intelligent control method, system and medium for automatic optical detection of appearance defects | |
| CN115439458A (en) | Industrial image defect target detection algorithm based on depth map attention | |
| CN115631472B (en) | Intelligent detection method for pedestrian intrusion on expressway | |
| CN111415339A (en) | Image defect detection method for complex texture industrial product | |
| CN114881998A (en) | Workpiece surface defect detection method and system based on deep learning | |
| CN115019294A (en) | Pointer instrument reading identification method and system | |
| CN110728269A (en) | High-speed rail contact net support pole number plate identification method | |
| CN106682604B (en) | Blurred image detection method based on deep learning | |
| CN117911796B (en) | Intelligent data processing system and method based on image recognition | |
| CN115984215A (en) | Fiber bundle defect detection method based on twin network | |
| CN115760808A (en) | Method, system and device for measuring size of plate glass and readable storage medium | |
| CN115200513A (en) | Coaxiality jumping detection, analysis and control system for rotating body clamp | |
| CN115601293A (en) | Object detection method and device, electronic equipment and readable storage medium | |
| CN112508946B (en) | Cable tunnel anomaly detection method based on antagonistic neural network | |
| CN114662594A (en) | Target feature recognition analysis system | |
| CN113220970A (en) | E-commerce big data platform based on block chain | |
| CN117523324B (en) | Image processing method and image sample classification method, device and storage medium | |
| CN117576098B (en) | Cell division balance evaluation method and device based on segmentation | |
| CN117654907B (en) | Automatic eliminating method and system for strip detector | |
| CN116523906B (en) | Method and system for detecting optical performance of glass substrate |
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 |