CN114627540A - Face information identification and comparison system for meeting reconnaissance application platform - Google Patents
Face information identification and comparison system for meeting reconnaissance application platform Download PDFInfo
- Publication number
- CN114627540A CN114627540A CN202210296892.9A CN202210296892A CN114627540A CN 114627540 A CN114627540 A CN 114627540A CN 202210296892 A CN202210296892 A CN 202210296892A CN 114627540 A CN114627540 A CN 114627540A
- Authority
- CN
- China
- Prior art keywords
- face
- image
- module
- feature
- recognition
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/25—Fusion techniques
- G06F18/253—Fusion techniques of extracted features
Landscapes
- Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Image Processing (AREA)
- Image Analysis (AREA)
Abstract
The invention relates to human face information recognition and comparison, in particular to a human face information recognition and comparison system for a reconnaissance application platform, which comprises a server and a human face recognition result output module, wherein the server receives videos and images to be recognized containing human face information on the reconnaissance application platform through a video image receiving module, controls a human face recognition model building module to build a human face recognition model, extracts a feature matrix from the images to be recognized through an image processing module, inputs the extracted feature matrix into the human face recognition model for model training and human face recognition through the human face recognition model training module, and outputs a human face recognition result of the human face recognition model through the human face recognition result output module; the technical scheme provided by the invention can effectively overcome the defects of low accuracy of face information identification and incapability of effectively identifying a single image in the prior art.
Description
Technical Field
The invention relates to face information identification and comparison, in particular to a face information identification and comparison system for a meeting investigation application platform.
Background
The on-site investigation and inspection of criminal cases is a investigation activity for the investigation and inspection of places, articles, persons and the like related to crimes by using scientific and technical means, and has the tasks of finding, fixing, extracting traces, material evidence and other information related to the crimes, storing related evidence data, analyzing the criminal process, judging the case properties, determining the investigation direction and range and providing clues and evidences for investigation and solution and criminal litigation.
The criminal investigation application platform is around criminal case data sharing, long-range investigation is examined, three directions of intelligent criminal investigation analysis, application big data intelligent analysis technique, block chain data storage technique, 3D scene simulation technique, 5G high-efficient data transmission technique and audio and video high definition imaging technique can effectively support the application of a plurality of scenes such as long-range investigation, on-the-spot instruction, volume evaluation, 3D scene restoration, information data analysis, result prejudgment. The development of the criminal meeting investigation application platform has important significance in the aspects of getting through a data fusion channel, breaking a data island, actively promoting the construction of a big data platform for communication and sharing, accelerating the formation of a data information resource service system covering the whole police and comprehensively utilizing the data information resource service system and the like.
After the video image of the face information of the suspect is collected, the investigation personnel can upload the video image to a criminal meeting investigation application platform to identify the face information. However, the accuracy of the existing face information recognition is low, and a single image cannot be effectively recognized, which brings great difficulty to the evidence collection work.
Disclosure of Invention
Technical problem to be solved
Aiming at the defects in the prior art, the invention provides a face information identification comparison system for a meeting survey application platform, which can effectively overcome the defects of low accuracy of face information identification and incapability of effectively identifying a single image in the prior art.
(II) technical scheme
In order to achieve the purpose, the invention is realized by the following technical scheme:
a face information recognition comparison system for a reconnaissance application platform comprises a server and a face recognition result output module, wherein the server receives videos and images to be recognized containing face information on the reconnaissance application platform through a video image receiving module, the server controls a face recognition model building module to build a face recognition model, the server extracts feature matrixes from the images to be recognized through an image processing module, the server inputs the extracted feature matrixes into the face recognition model through the face recognition model training module to perform model training and face recognition, and the face recognition result output module outputs face recognition results of the face recognition model;
the server extracts continuous frame images containing face information which cannot be identified by a face identification model from a video to be identified through a continuous frame image extraction module, the server compares face features in each frame image with collected face features in a face feature library through a face feature comparison module, the server merges the collected face features corresponding to each frame image through a face feature merging module to form a face feature merged set, the server fuses the face features in each frame image through a face feature fusion module to form a face feature fused set, the server calculates the similarity between the face feature merged set and the face feature fused set through a similarity calculation module, and the face identification result output module outputs a face identification result according to the similarity calculation result.
Preferably, the image processing module performs feature extraction on the image to be recognized by adopting a wavelet transform algorithm to obtain a transition feature matrix, and performs non-negative matrix decomposition on the intermediate feature matrix to obtain a recognition feature matrix.
Preferably, the image processing module performs feature extraction on the image to be recognized by using a wavelet transform algorithm to obtain a transition feature matrix, and the method includes:
the discrete wavelet transform algorithm is obtained by carrying out discretization processing on the continuous wavelet transform algorithm, and the image processing module carries out feature extraction on the image to be identified by adopting the discrete wavelet transform algorithm to obtain a transition feature matrix.
Preferably, the discretizing the continuous wavelet transform algorithm to obtain a discrete wavelet transform algorithm includes:
and respectively carrying out discretization processing on the scale factor and the translation factor in the continuous wavelet transform algorithm to obtain the discrete wavelet transform algorithm.
Preferably, the image to be recognized includes a training image, the image processing module extracts a recognition feature matrix from the training image and sends the recognition feature matrix to the face recognition model training module, and the face recognition model training module inputs the extracted recognition feature matrix into the face recognition model for model training to obtain an initial face recognition model.
Preferably, the face recognition model training module constructs a loss function, and continuously updates the weight parameters in the initial face recognition model until the loss function converges, so as to obtain the face recognition model.
Preferably, the human face feature fusion module fuses the human face features in each frame of image according to the weight of the human face features in each frame of image to form a human face feature fusion set;
and determining the weight of the human face features in each frame of image based on the blurring degree of each frame of image.
Preferably, after the continuous frame image extraction module extracts continuous frame images containing face information which cannot be recognized by a face recognition model from a video to be recognized, determining the fuzzy degree of each frame image based on a pre-trained image fuzzy degree judgment model;
the image blurring degree judging model is obtained by training a sample video frame image marked with a blurring value.
Preferably, the similarity calculation module calculates the similarity between the face feature union set and the face feature fusion set, and includes:
and calculating the similarity of the collected face features in the face feature merging set of each frame of image based on the face feature merging set, and sequencing the similarity.
Preferably, the face recognition result output module outputs the face recognition result according to the similarity calculation result, and includes:
and the face recognition result output module traces the similarity sequence of each frame of image, acquires face information corresponding to the collected face features in the face feature library in the similarity sequence of each frame of image, matches the most similar face information in the face feature library according to the average calculation result of the similarity, and outputs the face information as a face recognition result.
(III) advantageous effects
Compared with the prior art, the face information identification and comparison system for the meeting survey application platform provided by the invention has the following beneficial effects:
1) the video image receiving module receives an image to be recognized containing face information on a meeting survey inspection application platform, the image processing module extracts features of the image to be recognized by adopting a wavelet transform algorithm to obtain a transition feature matrix, and carries out non-negative matrix decomposition on an intermediate feature matrix to obtain a recognition feature matrix, and the face recognition model carries out face recognition on the extracted recognition feature matrix, so that the accuracy of carrying out face information recognition on a single-frame image can be effectively improved;
2) the face feature comparison module compares the face features in each frame of image with the collected face features in the face feature library, the face feature merging module merges the collected face features corresponding to each frame of image to form a face feature merged set, the face feature fusion module fuses the face features in each frame of image to form a face feature fusion set, the similarity calculation module calculates the similarity between the face feature merged set and the face feature fusion set and outputs a face recognition result according to the similarity calculation result, so that the accuracy of face information recognition can be improved by using continuous frame images under the condition that a single frame of image cannot be effectively recognized.
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. It is obvious that the drawings in the following description are only some embodiments of the invention, and that for a person skilled in the art, other drawings can be derived from them without inventive effort.
FIG. 1 is a schematic diagram of the system of the present invention;
fig. 2 is a schematic flow chart of face information recognition of continuous frame images in the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the embodiments of the present invention clearer, 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. It is to be understood that the embodiments described are only a few embodiments of the present invention, and not all 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.
A face information identification comparison system for an reconnaissance and reconnaissance application platform is shown in figure 1 and comprises a server and a face identification result output module, wherein the server receives videos to be identified and images containing face information on the reconnaissance and reconnaissance application platform through a video image receiving module, the server controls a face identification model building module to build a face identification model, the server extracts feature matrixes from the images to be identified through an image processing module, the server inputs the extracted feature matrixes into the face identification model through the face identification model training module to perform model training and face identification, and the face identification result output module outputs face identification results of the face identification model.
Firstly, the face recognition model training module needs to perform model training on the face recognition model constructed by the face recognition model construction module, and after the initial face recognition model is obtained through training, model optimization is performed on the initial face recognition model, and finally the face recognition model after training optimization is obtained, and the specific process is as follows:
the image to be recognized comprises a training image, the image processing module extracts a recognition characteristic matrix from the training image and sends the recognition characteristic matrix to the face recognition model training module, and the face recognition model training module inputs the extracted recognition characteristic matrix into a face recognition model for model training to obtain an initial face recognition model;
and the face recognition model training module constructs a loss function, and continuously updates the weight parameters in the initial face recognition model until the loss function is converged to obtain the face recognition model.
The process of extracting the characteristic matrix from the image to be identified by the image processing module is as follows:
the image processing module extracts the features of the image to be recognized by adopting a wavelet transform algorithm to obtain a transition feature matrix, and performs non-negative matrix decomposition on the intermediate feature matrix to obtain a recognition feature matrix.
The image processing module adopts a wavelet transform algorithm to perform feature extraction on an image to be identified to obtain a transition feature matrix, and the method comprises the following steps:
the image processing module adopts the discrete wavelet transform algorithm to extract the characteristics of the image to be identified so as to obtain a transition characteristic matrix.
According to the technical scheme, the video image receiving module receives the image to be recognized containing face information on the meeting survey application platform, the image processing module extracts the features of the image to be recognized by adopting a wavelet transform algorithm to obtain a transition feature matrix, non-negative matrix decomposition is carried out on the middle feature matrix to obtain a recognition feature matrix, and the face recognition model carries out face recognition on the extracted recognition feature matrix, so that the accuracy of face information recognition on a single-frame image can be effectively improved.
As shown in fig. 1 and 2, a server extracts continuous frame images including face information that cannot be recognized by a face recognition model from a video to be recognized through a continuous frame image extraction module, the server compares face features in each frame image with collected face features in a face feature library through a face feature comparison module, the server combines the collected face features corresponding to each frame image through a face feature combination module to form a face feature combination set, the server combines the face features in each frame image through a face feature combination module to form a face feature combination set, the server calculates the similarity between the face feature combination set and the face feature combination set through a similarity calculation module, and a face recognition result output module outputs a face recognition result according to a similarity calculation result.
And after extracting the continuous frame images containing the face information which cannot be identified by the face identification model from the video to be identified, the continuous frame image extraction module determines the fuzzy degree of each frame image based on a pre-trained image fuzzy degree judgment model. The image blurring degree judging model is obtained by training a sample video frame image marked with a blurring value.
And the human face feature fusion module fuses the human face features in each frame of image according to the weight of the human face features in each frame of image to form a human face feature fusion set. The weight of the face features in each frame image is determined based on the blurring degree of each frame image, and the higher the blurring degree of the image is, the smaller the weight of the face features in the frame image is.
The similarity calculation module calculates the similarity between the face feature combination set and the face feature fusion set, and comprises the following steps:
and calculating similarity of collected face features in the face feature merging set of each frame of image based on the face feature merging set, and sequencing the similarity.
The face recognition result output module outputs a face recognition result according to the similarity calculation result, and comprises:
and the face recognition result output module traces the similarity sequence of each frame of image, acquires face information corresponding to the collected face features in the face feature library in the similarity sequence of each frame of image, matches the most similar face information in the face feature library according to the average calculation result of the similarity, and outputs the face information as a face recognition result.
In the technical scheme, when the single-frame image cannot be effectively identified due to high fuzzy degree of the single-frame image, the continuous frame image extraction module extracts the continuous frame image containing the information of the face which cannot be effectively identified from the video to be identified. The human face feature comparison module compares the human face features in each frame of image with the collected human face features in the human face feature library, and the human face feature merging module merges the collected human face features corresponding to each frame of image to form a human face feature merging set; and the human face feature fusion module fuses the human face features in each frame of image to form a human face feature fusion set.
The similarity calculation module calculates similarity of collected face features in the face feature merging set of each frame of image based on the face feature merging set, and carries out similarity sequencing; and the face recognition result output module traces the similarity sequence of each frame of image, acquires face information corresponding to the collected face features in the face feature library in the similarity sequence of each frame of image, matches the most similar face information in the face feature library according to the average calculation result of the similarity, and outputs the face information as a face recognition result. Therefore, the accuracy of face information recognition can be improved by using continuous frame images under the condition that single frame images cannot be effectively recognized.
The above examples are only intended to illustrate the technical solution of the present invention, and not to limit it; although the present invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; such modifications and substitutions do not depart from the spirit and scope of the corresponding technical solutions.
Claims (10)
1. A face information identification and comparison system for a meeting reconnaissance application platform is characterized in that: the system comprises a server and a face recognition result output module, wherein the server receives videos and images to be recognized containing face information on a reconnaissance inspection application platform through a video image receiving module, the server controls a face recognition model construction module to construct a face recognition model, the server extracts a feature matrix from the images to be recognized through an image processing module, the server inputs the extracted feature matrix into the face recognition model through a face recognition model training module to perform model training and face recognition, and the face recognition result output module outputs a face recognition result of the face recognition model;
the server extracts continuous frame images containing face information which cannot be identified by a face identification model from a video to be identified through a continuous frame image extraction module, the server compares face features in each frame image with collected face features in a face feature library through a face feature comparison module, the server merges the collected face features corresponding to each frame image through a face feature merging module to form a face feature merged set, the server fuses the face features in each frame image through a face feature fusion module to form a face feature fused set, the server calculates the similarity between the face feature merged set and the face feature fused set through a similarity calculation module, and the face identification result output module outputs a face identification result according to the similarity calculation result.
2. The system of claim 1, wherein the face information recognition and comparison system comprises: the image processing module extracts the features of the image to be recognized by adopting a wavelet transform algorithm to obtain a transition feature matrix, and performs non-negative matrix decomposition on the intermediate feature matrix to obtain a recognition feature matrix.
3. The system of claim 2, wherein the face information recognition and comparison system comprises: the image processing module adopts a wavelet transformation algorithm to perform feature extraction on an image to be identified to obtain a transition feature matrix, and the method comprises the following steps:
the discrete wavelet transform algorithm is obtained by carrying out discretization processing on the continuous wavelet transform algorithm, and the image processing module carries out feature extraction on the image to be identified by adopting the discrete wavelet transform algorithm to obtain a transition feature matrix.
4. The system of claim 3, wherein the face information recognition and comparison system comprises: the discretization processing of the continuous wavelet transform algorithm to obtain the discrete wavelet transform algorithm comprises the following steps:
and respectively carrying out discretization processing on the scale factor and the translation factor in the continuous wavelet transform algorithm to obtain the discrete wavelet transform algorithm.
5. The system of claim 2, wherein the face information recognition and comparison system comprises: the image to be recognized comprises a training image, the image processing module extracts a recognition feature matrix from the training image and sends the recognition feature matrix to the face recognition model training module, and the face recognition model training module inputs the extracted recognition feature matrix into a face recognition model for model training to obtain an initial face recognition model.
6. The system of claim 5, wherein the face information recognition and comparison system comprises: and the face recognition model training module constructs a loss function, and continuously updates the weight parameters in the initial face recognition model until the loss function is converged to obtain the face recognition model.
7. The system of claim 1, wherein the face information recognition and comparison system comprises: the human face feature fusion module fuses the human face features in each frame of image according to the weight of the human face features in each frame of image to form a human face feature fusion set;
and determining the weight of the human face features in each frame of image based on the blurring degree of each frame of image.
8. The system of claim 7, wherein the face information recognition and comparison system comprises: the continuous frame image extraction module extracts continuous frame images which contain face information which cannot be identified by a face identification model from a video to be identified, and then determines the fuzzy degree of each frame image based on a pre-trained image fuzzy degree judgment model;
the image blurring degree judging model is obtained by training a sample video frame image marked with a blurring value.
9. The system of claim 7, wherein the face information recognition and comparison system comprises: the similarity calculation module calculates the similarity between the human face feature fusion set and the human face feature fusion set, and comprises the following steps:
and calculating similarity of collected face features in the face feature merging set of each frame of image based on the face feature merging set, and sequencing the similarity.
10. The system of claim 9, wherein the face information recognition and comparison system comprises: the face recognition result output module outputs a face recognition result according to the similarity calculation result, and comprises:
and the face recognition result output module traces the similarity sequence of each frame of image, acquires face information corresponding to the collected face features in the face feature library in the similarity sequence of each frame of image, matches the most similar face information in the face feature library according to the average calculation result of the similarity, and outputs the face information as a face recognition result.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210296892.9A CN114627540A (en) | 2022-03-24 | 2022-03-24 | Face information identification and comparison system for meeting reconnaissance application platform |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210296892.9A CN114627540A (en) | 2022-03-24 | 2022-03-24 | Face information identification and comparison system for meeting reconnaissance application platform |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114627540A true CN114627540A (en) | 2022-06-14 |
Family
ID=81903984
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210296892.9A Pending CN114627540A (en) | 2022-03-24 | 2022-03-24 | Face information identification and comparison system for meeting reconnaissance application platform |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114627540A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117528131A (en) * | 2024-01-05 | 2024-02-06 | 青岛美迪康数字工程有限公司 | AI integrated display system and method for medical image |
-
2022
- 2022-03-24 CN CN202210296892.9A patent/CN114627540A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117528131A (en) * | 2024-01-05 | 2024-02-06 | 青岛美迪康数字工程有限公司 | AI integrated display system and method for medical image |
| CN117528131B (en) * | 2024-01-05 | 2024-04-05 | 青岛美迪康数字工程有限公司 | AI integrated display system and method for medical image |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109919031B (en) | Human behavior recognition method based on deep neural network | |
| CN111798456A (en) | Instance segmentation model training method and device and instance segmentation method | |
| CN111054080B (en) | Method, device and equipment for intelligently detecting perspective plug-in and storage medium thereof | |
| CN112699786B (en) | Video behavior identification method and system based on space enhancement module | |
| CN113284144B (en) | Tunnel detection method and device based on unmanned aerial vehicle | |
| CN112560827B (en) | Model training method, model training device, model prediction method, electronic device, and medium | |
| CN113963315A (en) | Real-time video multi-user behavior recognition method and system in complex scene | |
| CN110390308B (en) | Video behavior identification method based on space-time confrontation generation network | |
| CN115719475B (en) | Three-stage trackside equipment fault automatic detection method based on deep learning | |
| CN113569672A (en) | Lightweight target detection and fault identification method, device and system | |
| CN114998830A (en) | Wearing detection method and system for safety helmet of transformer substation personnel | |
| CN115147383A (en) | Insulator state rapid detection method based on lightweight YOLOv5 model | |
| CN109670423A (en) | A kind of image identification system based on deep learning, method and medium | |
| CN116994021A (en) | Image detection method, device, computer readable medium and electronic equipment | |
| CN114627540A (en) | Face information identification and comparison system for meeting reconnaissance application platform | |
| Miao et al. | Abnormal behavior learning based on edge computing toward a crowd monitoring system | |
| CN113191274A (en) | Oil field video intelligent safety event detection method and system based on neural network | |
| CN115661904A (en) | Data labeling and domain adaptation model training method, device, equipment and medium | |
| CN114898181A (en) | Hidden danger violation identification method and device for explosion-related video | |
| CN114445875A (en) | Deep learning-based identity recognition and face comparison system and training method | |
| CN115100546A (en) | Mobile-based small target defect identification method and system for power equipment | |
| CN111046785A (en) | Method for identifying key target of unmanned aerial vehicle routing inspection video based on convolutional neural network | |
| CN110728217A (en) | SAR image recognition method, device, equipment and storage medium | |
| Nazarkevych et al. | A YOLO-based Method for Object Contour Detection and Recognition in Video Sequences. | |
| CN117152668B (en) | Intelligent logistics implementation method, device and equipment based on Internet of things |
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 |