CN114529970A - Pedestrian detection system based on fusion of Gabor features and HOG features - Google Patents
Pedestrian detection system based on fusion of Gabor features and HOG features Download PDFInfo
- Publication number
- CN114529970A CN114529970A CN202210147839.2A CN202210147839A CN114529970A CN 114529970 A CN114529970 A CN 114529970A CN 202210147839 A CN202210147839 A CN 202210147839A CN 114529970 A CN114529970 A CN 114529970A
- Authority
- CN
- China
- Prior art keywords
- gabor
- feature
- image
- features
- hog
- 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
- G06F18/213—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
- G06F18/2135—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods based on approximation criteria, e.g. principal component analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
-
- 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)
- Artificial Intelligence (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (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 Analysis (AREA)
Abstract
The invention relates to the technical field of intelligent monitoring, and discloses a pedestrian detection system based on fusion of Gabor features and HOG features. According to the invention, through calculating the gradient of each pixel of each unit, counting the gradient direction histogram of the cell units, and dividing 6 bins at intervals of 30 degrees, each cell unit can obtain a 6-dimensional feature vector, and finally, the feature vectors of block normalization are connected in series to synthesize HOG feature vectors.
Description
Technical Field
The invention relates to the technical field of intelligent monitoring, in particular to a pedestrian detection system based on the fusion of Gabor characteristics and HOG characteristics.
Background
The face recognition technology is a technology for performing identity authentication by using face features, and belongs to a biometric feature recognition technology as well as an iris recognition technology and a fingerprint recognition technology. But compared with the iris and fingerprint characteristics, the human face characteristics not only have uniqueness and persistence, but also can carry out identity authentication without contact, so that the human face identification technology gradually becomes the mainstream of an identity authentication mode, a large number of products based on human face identification are also successively fallen to the ground, and the life of people is greatly facilitated.
The Gabor feature extraction algorithm can extract detail texture features of the face image in different directions, has certain robustness on illumination, shielding and expression change, and the HOG feature focuses on edge features of the face image. The Gabor feature and the HOG feature are fused, and a fused feature (referred to as G-H feature) extraction algorithm is designed. The dimension of the fusion feature obtained by the fusion feature extraction algorithm is too large, which is not beneficial to calculation, the PCA is adopted to reduce the dimension of the fusion feature, and finally the SVM is used as a classifier to recognize the face.
Disclosure of Invention
The invention aims to provide a pedestrian detection system based on Gabor feature and HOG feature fusion, which achieves the purposes of reducing the dimension of fusion features by adopting PCA and finally performing face recognition by using an SVM as a classifier, simplifies the operation and improves the face recognition efficiency.
In order to achieve the purpose, the invention provides the following technical scheme:
a pedestrian detection system based on fusion of Gabor features and HOG features comprises the following steps:
s1, preprocessing the input image, if the image is not a face image, performing face detection, face screenshot and face alignment on the image, and performing graying, filtering and normalization on the face image.
And S2, constructing a Gabor filter bank, setting the kernel size and the kernel direction of the filter, and constructing the Gabor filter bank.
And S3, carrying out Gabor transformation on the normalized face image to obtain Gabor characteristics, wherein the Gabor characteristics are 20 Gabor characteristic graphs, and the characteristic graphs are primary characteristics.
S4, extracting HOG features from a Gabor feature map with the size of 96X112, eliminating the influence of uneven illumination on the image by Gamma correction, calculating gradients in the horizontal direction and the vertical direction aiming at each pixel point (X, y) in the image, and dividing the image into a plurality of regions called as cell units.
And S5, compressing the secondary feature map into one-dimensional feature vectors, and connecting the feature vectors in series to form the fusion features.
And S6, performing dimensionality reduction on the fusion features by using principal component analysis to obtain data features subjected to dimensionality reduction, classifying the G-H data subjected to dimensionality reduction by using an SVM (support vector machine), and finally outputting a result.
Preferably, in S2, the filter kernel size is set to 3, 5, 7, 9, and 11, the direction is set to 0 °, 45 °, 90 °, and 135 °, and one Gabor filter group is generated, where 4X5 ═ 20 different Gabor filters can be obtained.
Preferably, in S4, 96X112 is divided into 12X14 cell units, each cell unit has a size of 8X8, and every 2X2 cell units are divided into one block, which can be divided into 6X7 blocks.
Preferably, in S4, the gradient of each pixel of each cell is calculated, the histogram of gradient direction of the cell unit is counted, 6 bins are divided at intervals of 30 °, so that each cell unit can obtain a 6-dimensional feature vector, and finally, the feature vectors normalized by the blocks are concatenated to synthesize the HOG feature vector, so as to obtain 20 feature maps, which are secondary features.
The invention provides a pedestrian detection system based on the fusion of Gabor characteristics and HOG characteristics, which has the following beneficial effects:
according to the invention, through calculating the gradient of each pixel of each unit, counting the gradient direction histogram of the cell units, and dividing 6 bins at intervals of 30 degrees, each cell unit can obtain a 6-dimensional feature vector, and finally, the feature vectors of block normalization are connected in series to synthesize HOG feature vectors.
Drawings
Fig. 1 is a schematic process diagram of a pedestrian detection system based on the fusion of Gabor features and HOG features.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
As shown in fig. 1, the present invention provides a technical solution:
a pedestrian detection system based on fusion of Gabor features and HOG features comprises the following steps:
s1, preprocessing the input image, if the image is not a face image, performing face detection, face screenshot and face alignment on the image, and performing graying, filtering and normalization on the face image.
And S2, constructing a Gabor filter bank, setting the kernel size and the kernel direction of the filter, and constructing the Gabor filter bank.
And S3, carrying out Gabor transformation on the normalized face image to obtain Gabor characteristics, wherein the Gabor characteristics are 20 Gabor characteristic graphs, and the characteristic graphs are primary characteristics.
S4, extracting HOG features from a Gabor feature map with the size of 96X112, eliminating the influence of uneven illumination on the image by Gamma correction, calculating gradients in the horizontal direction and the vertical direction aiming at each pixel point (X, y) in the image, and dividing the image into a plurality of regions called as cell units.
And S5, compressing the secondary feature map into one-dimensional feature vectors, and connecting the feature vectors in series to form the fusion features.
And S6, performing dimensionality reduction on the fusion features by using principal component analysis to obtain data features subjected to dimensionality reduction, classifying the G-H data subjected to dimensionality reduction by using an SVM (support vector machine), and finally outputting a result.
In summary, the gradient of each pixel of each unit is calculated, the histogram of the gradient direction of the cell unit is counted, 6 bins are divided at an interval of 30 degrees, each cell unit can obtain a 6-dimensional feature vector, finally, the feature vectors of block normalization are connected in series to synthesize an HOG feature vector, through the process, 20 feature maps can be obtained, the feature maps are two-level features, PCA is adopted to reduce the dimension of the fusion features, and SVM is finally used as a classifier to perform face recognition, so that the operation is simplified, and the face recognition efficiency is improved.
It is noted that, herein, relational terms such as first and second, and the like may be 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. Also, 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.
Claims (4)
1. Pedestrian detection system based on fusion of Gabor characteristics and HOG characteristics is characterized by comprising the following steps:
s1, firstly preprocessing an input image, if the image is not a face image, firstly carrying out face detection, face screenshot and face alignment on the image, and then carrying out graying, filtering and normalization on the face image;
s2, constructing a Gabor filter bank, setting the kernel size and the kernel direction of the filter, and constructing the Gabor filter bank;
s3, carrying out Gabor transformation on the normalized face image to obtain Gabor characteristics, wherein the Gabor characteristics are 20 Gabor characteristic graphs, and the characteristic graphs are primary characteristics;
s4, extracting HOG characteristics from a Gabor characteristic diagram with the size of 96X112, eliminating the influence of uneven illumination on the image by adopting Gamma correction, calculating gradients in the horizontal direction and the vertical direction aiming at each pixel point (X, y) in the image, and dividing the image into a plurality of regions called as cell units;
s5, compressing the secondary feature map into one-dimensional feature vectors, and connecting the feature vectors in series to form fusion features;
and S6, performing dimensionality reduction on the fusion features by using principal component analysis to obtain data features subjected to dimensionality reduction, classifying the G-H data subjected to dimensionality reduction by using an SVM (support vector machine), and finally outputting a result.
2. The Gabor feature and HOG feature fusion based pedestrian detection system according to claim 1, wherein: in S2, the filter kernel sizes are set to 3, 5, 7, 9, and 11, and the directions are set to 0 °, 45 °, 90 °, and 135 °, so as to generate one Gabor filter bank, where 4X5 ═ 20 different Gabor filters can be obtained.
3. The Gabor feature and HOG feature fusion based pedestrian detection system according to claim 1, wherein: in S4, 96X112 is divided into 12X14 cell units each having a size of 8X8, and each 2X2 cell units are divided into one block, and 6X7 blocks can be divided.
4. The Gabor feature and HOG feature fusion based pedestrian detection system according to claim 3, wherein: in S4, the gradient of each pixel of each cell is calculated, the histogram of the gradient direction of the cell is counted, 6 bins are divided at intervals of 30 °, each cell can obtain a 6-dimensional feature vector, and finally, the feature vectors normalized by the blocks are concatenated to synthesize an HOG feature vector, so that 20 feature maps, which are secondary features, can be obtained.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210147839.2A CN114529970A (en) | 2022-02-17 | 2022-02-17 | Pedestrian detection system based on fusion of Gabor features and HOG features |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210147839.2A CN114529970A (en) | 2022-02-17 | 2022-02-17 | Pedestrian detection system based on fusion of Gabor features and HOG features |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114529970A true CN114529970A (en) | 2022-05-24 |
Family
ID=81622491
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210147839.2A Pending CN114529970A (en) | 2022-02-17 | 2022-02-17 | Pedestrian detection system based on fusion of Gabor features and HOG features |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114529970A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115563478A (en) * | 2022-12-05 | 2023-01-03 | 电子科技大学 | Millimeter wave radar non-line-of-sight human behavior recognition system based on multi-class feature fusion |
-
2022
- 2022-02-17 CN CN202210147839.2A patent/CN114529970A/en active Pending
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115563478A (en) * | 2022-12-05 | 2023-01-03 | 电子科技大学 | Millimeter wave radar non-line-of-sight human behavior recognition system based on multi-class feature fusion |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Yuan et al. | Fingerprint liveness detection using an improved CNN with image scale equalization | |
| CN101923640B (en) | Method for distinguishing false iris images based on robust texture features and machine learning | |
| CN107135664A (en) | The method and face identification device of a kind of recognition of face | |
| Nhat et al. | Feature fusion by using LBP, HOG, GIST descriptors and Canonical Correlation Analysis for face recognition | |
| CN107392142B (en) | Method and device for identifying true and false face | |
| CN102156887A (en) | Human face recognition method based on local feature learning | |
| Sudha et al. | Comparative study of features fusion techniques | |
| CN102902980A (en) | Linear programming model based method for analyzing and identifying biological characteristic images | |
| Lee et al. | A Gabor filter-based approach to fingerprint recognition | |
| Naderi et al. | Fusing iris, palmprint and fingerprint in a multi-biometric recognition system | |
| CN104881635A (en) | Image texture extraction and identification method by non-Gauss two-dimension Gabor filter | |
| CN110826444A (en) | Facial expression recognition method and system based on Gabor filter | |
| Ramya et al. | Certain investigation on iris image recognition using hybrid approach of Fourier transform and Bernstein polynomials | |
| CN114529970A (en) | Pedestrian detection system based on fusion of Gabor features and HOG features | |
| Benzaoui et al. | Face recognition using 1DLBP, DWT and SVM | |
| Yuan et al. | A review of recent advances in ear recognition | |
| Chergui et al. | Kinship verification using BSIF and LBP | |
| Kasaer et al. | Fingerprint feature enhancement using block-direction on reconstructed images | |
| CN116311467A (en) | Identity verification processing method and device based on face recognition | |
| Karungaru et al. | Face recognition in colour images using neural networks and genetic algorithms | |
| Rashid et al. | Face recognition system based on gabor wavelets transform, principal component analysis and support vector machine | |
| Wang et al. | Face recognition using Gaborface-based 2DPCA and (2D) 2PCA classification with ensemble and multichannel model | |
| CN111191624A (en) | Finger vein feature extraction algorithm based on normal vector information | |
| Acula et al. | Offline Signature Verification System using CNN Algorithm Combined with Histogram of Oriented Gradients | |
| Chen et al. | Classification of smile expression using hybrid phog and gabor features |
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 |