CN115393585A - Moving target detection method based on super-pixel fusion network - Google Patents
Moving target detection method based on super-pixel fusion network Download PDFInfo
- Publication number
- CN115393585A CN115393585A CN202210962818.6A CN202210962818A CN115393585A CN 115393585 A CN115393585 A CN 115393585A CN 202210962818 A CN202210962818 A CN 202210962818A CN 115393585 A CN115393585 A CN 115393585A
- Authority
- CN
- China
- Prior art keywords
- layer
- pixel
- super
- image
- histogram
- 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.)
- Granted
Links
- 238000001514 detection method Methods 0.000 title claims abstract description 30
- 230000004927 fusion Effects 0.000 title claims abstract description 16
- 238000000034 method Methods 0.000 claims abstract description 7
- 230000004913 activation Effects 0.000 claims description 27
- 238000010606 normalization Methods 0.000 claims description 27
- 238000011176 pooling Methods 0.000 claims description 27
- 238000001914 filtration Methods 0.000 claims description 8
- 238000012549 training Methods 0.000 claims description 8
- 230000011218 segmentation Effects 0.000 claims description 7
- 238000012935 Averaging Methods 0.000 claims description 6
- 238000013528 artificial neural network Methods 0.000 claims description 3
- 238000013527 convolutional neural network Methods 0.000 claims description 3
- 230000004807 localization Effects 0.000 claims description 3
- 238000012545 processing Methods 0.000 description 3
- 230000000694 effects Effects 0.000 description 2
- 239000000284 extract Substances 0.000 description 2
- 239000011159 matrix material Substances 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 230000000877 morphologic effect Effects 0.000 description 1
- 238000012805 post-processing Methods 0.000 description 1
- 238000007619 statistical method Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/26—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/11—Region-based segmentation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/36—Applying a local operator, i.e. means to operate on image points situated in the vicinity of a given point; Non-linear local filtering operations, e.g. median filtering
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/50—Extraction of image or video features by performing operations within image blocks; by using histograms, e.g. histogram of oriented gradients [HoG]; by summing image-intensity values; Projection analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
- G06V10/774—Generating sets of training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
- G06V10/80—Fusion, i.e. combining data from various sources at the sensor level, preprocessing level, feature extraction level or classification level
- G06V10/806—Fusion, i.e. combining data from various sources at the sensor level, preprocessing level, feature extraction level or classification level of extracted features
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/82—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Computation (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Computing Systems (AREA)
- Databases & Information Systems (AREA)
- General Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Nonlinear Science (AREA)
- Image Analysis (AREA)
Abstract
The invention relates to the technical field of target detection, in particular to a moving target detection method based on a superpixel fusion network. And then, extracting histogram features of the candidate foreground superpixels, and then respectively taking the pixel features and the superpixel features as the input of a finding pixel fusion network, wherein the whole process has high operation speed and strong robustness.
Description
Technical Field
The invention relates to the technical field of target detection, in particular to a moving target detection method based on a super-pixel fusion network.
Background
The moving object detection is one of applications of image processing, generally speaking, a background model is obtained through a statistical method, the background model is updated in real time to adapt to light changes and changes of a scene, a morphological method and a detection connected domain area are used for post-processing, influences caused by noise and background disturbance are eliminated, shadows are detected in an HSV (hue, saturation, value) chromaticity space, an accurate moving object is obtained, in a complex scene, the moving object detection is still a challenging task, an existing method based on deep learning mainly adopts a u-net network and obtains a surprising effect, however, the local continuity between pixels is ignored, the detection performance is required to be further improved, in addition, the network contains information of the scene, and the generalization capability is required to be further improved.
Disclosure of Invention
In view of the above situation, an object of the present invention is to provide a moving object detection method based on a super-pixel fusion network.
The technical purpose of the invention is realized by the following technical scheme:
a moving target detection method based on a super-pixel fusion network comprises two parts: (1) a training stage; (2) a detection stage;
the detection phase comprises:
step 1, inputting a color image sequence R1, averaging 3 channel numerical values to perform image graying, and obtaining a grayed image sequence G1;
step 2, performing median filtering on the image sequence to obtain a background image B1, and performing difference on the image sequence G1 and the background image B1 to obtain a candidate foreground sequence which is marked as a pixel characteristic F1;
step 3, performing superpixel segmentation on the color image sequence R1 to obtain region information C1;
step 4, calculating a histogram of pixels of a candidate foreground region corresponding to the superpixel according to the region information, wherein the interval is 0.1 when the histogram is in a range of [ -1,1 ];
step 5, taking the histogram of each region as the characteristics of all pixels in the region, and recording the histogram as a super-pixel characteristic F2;
step 6, constructing a network;
step 7, training the model;
step 8, outputting the trained network model M;
the detection phase comprises:
step 9, inputting an image sequence R2, and averaging 3-channel numerical values to perform image graying if the image is a color image to obtain a grayed image sequence G2; if the image is a gray image, directly enabling G2= R2;
step 10, performing median filtering on the image sequence to obtain a background image B2, and performing difference on the image sequence G2 and the background image B2 to obtain a candidate foreground sequence which is recorded as a pixel characteristic F3;
step 11, performing superpixel segmentation on the color image sequence R2 to obtain region information C2;
step 12, calculating a histogram of pixels of a candidate foreground region corresponding to the super-pixel according to the region information, wherein the interval is 0.1 when the histogram is in a range of [ -1,1 ];
step 13, taking the histogram of each region as the characteristics of all pixels in the region, and recording the histogram as a super-pixel characteristic F4;
step 14, taking the super pixel characteristics F4 and the pixel characteristics F3 as the input of the trained network model M;
and step 15, outputting a detection result.
Further, the specific method of step 6 is as follows:
(1) constructing an encoder:
the convolutional neural network comprises an input layer, a hidden layer and an output layer;
the input layer comprises two inputs, the resolution is 240 multiplied by 320, the number of channels of the encoder input corresponding to the super pixel characteristic F2 is 21, the encoder input corresponding to the pixel characteristic F1 is 1, and the convolution size in the convolution neural network is 3 multiplied by 3;
layer 1 of the hidden layers adopts convolution, batch normalization, an activation layer and a pooling layer Conv + BN + Relu + Maxpool, and 8 convolutions are used for generating 8 feature maps;
layer 2 of the hidden layers adopts convolution, batch normalization, an activation layer and a pooling layer Conv + BN + Relu + Maxpool, and 16 convolutions are used for generating 16 feature maps;
the 3 rd layer in the hidden layers adopts convolution, batch normalization, an activation layer and a pooling layer Conv + BN + Relu + Maxpool, and 32 convolutions are used for generating 32 feature maps;
the 4 th layer in the hidden layers adopts convolution, batch normalization, an activation layer and a pooling layer Conv + BN + Relu + Maxpool, and 64 feature maps are generated by using 64 convolutions;
(2) constructing a connecting layer:
the 5 th layer in the hidden layers is a connection layer, and the connection layer connects the two encoders by using localization;
(3) constructing a decoder:
layer 6 of the hidden layers adopts convolution, batch normalization, an activation layer and a pooling layer Conv + BN + Relu, and uses 128 convolutions to generate 64 feature maps;
the 7 th layer in the hidden layer adopts convolution, batch normalization, an activation layer and a pooling layer Deconv + BN + Relu, and generates 32 feature maps by using 64 convolutions;
the 8 th layer in the hidden layers adopts convolution, batch normalization, an activation layer and a pooling layer Deconv + BN + Relu, and 32 convolutions are used for generating 16 feature maps;
the 9 th layer in the hidden layers adopts convolution, batch normalization, an activation layer and a pooling layer Deconv + BN + Relu, and 8 convolutions are used for generating 8 feature maps;
the 10 th layer in the hidden layer adopts convolution, batch normalization, an activation layer and a pooling layer Deconv + BN + ClippedRelu, and 1 feature map is generated by using 1 convolution;
the output layer comprises a regression layer;
and taking the super pixel characteristics and the pixel characteristics as the input of the network, and outputting the super pixel characteristics and the pixel characteristics as the group of the corresponding input image.
In conclusion, the invention has the following beneficial effects:
the invention firstly uses median filtering to obtain candidate foreground, then judges whether the pixel is a foreground pixel or not through a super-pixel fusion network, and only relates to simple multiplication of a matrix when detecting, so the invention has small time complexity, and the processing speed of a training stage and a detection stage is high.
Drawings
The accompanying drawings, which are included to provide a further understanding of the invention and are incorporated in and constitute a part of this application, are not intended to limit the invention, and:
FIG. 1 is a schematic diagram of the steps of the present invention.
Detailed Description
The foregoing and other technical and scientific aspects, features and utilities of the present invention will be apparent from the following detailed description of the embodiments, which is to be read in connection with the accompanying drawings of fig. 1. The structural contents mentioned in the following embodiments are all referred to the attached drawings of the specification.
Exemplary embodiments of the present invention will be described below with reference to the accompanying drawings.
Example 1: a moving target detection method based on a super-pixel fusion network comprises two parts: (1) a training stage; (2) a detection stage;
the detection phase comprises:
step 1, inputting a color image sequence R1, averaging 3-channel numerical values to perform image graying, and obtaining a grayed image sequence G1;
step 2, performing median filtering on the image sequence to obtain a background image B1, and performing difference on the image sequence G1 and the background image B1 to obtain a candidate foreground sequence which is marked as a pixel characteristic F1;
step 3, performing superpixel segmentation on the color image sequence R1 to obtain region information C1;
step 4, calculating a histogram of pixels of a candidate foreground region corresponding to the superpixel according to the region information, wherein the interval is 0.1 when the histogram is in a range of [ -1,1 ];
step 5, taking the histogram of each region as the characteristics of all pixels in the region, and recording the histogram of each region as a super-pixel characteristic F2;
step 6, constructing a network:
(1) constructing an encoder:
the convolutional neural network comprises an input layer, a hidden layer and an output layer;
the input layer comprises two inputs, the resolution is 240 multiplied by 320, the number of channels of the encoder input corresponding to the super pixel characteristic F2 is 21, the encoder input corresponding to the pixel characteristic F1 is 1, and the convolution size in the convolution neural network is 3 multiplied by 3;
the layer 1 in the hidden layer adopts convolution, batch normalization, an activation layer and a pooling layer Conv + BN + Relu + Maxpool, and 8 feature maps are generated by using 8 convolutions;
the layer 2 in the hidden layer adopts convolution, batch normalization, an activation layer and a pooling layer Conv + BN + Relu + Maxpool, and 16 feature maps are generated by using 16 convolutions;
the 3 rd layer in the hidden layer adopts convolution, batch normalization, an activation layer and a pooling layer Conv + BN + Relu + Maxpool, and 32 feature maps are generated by using 32 convolutions;
layer 4 of the hidden layer adopts convolution, batch normalization, an activation layer and a pooling layer Conv + BN + Relu + Maxpool, and 64 feature maps are generated by using 64 convolutions;
(2) constructing a connecting layer:
the 5 th layer in the hidden layer is a connection layer, and the connection layer connects the two encoders by using localization;
(3) constructing a decoder:
layer 6 of the hidden layers employs convolution, batch normalization, an activation layer and a pooling layer Conv + BN + Relu, and 128 convolutions are used to generate 64 feature maps;
the 7 th layer in the hidden layer adopts convolution, batch normalization, an activation layer and a pooling layer Deconv + BN + Relu, and 64 convolutions are used for generating 32 feature maps;
the 8 th layer in the hidden layer adopts convolution, batch normalization, an activation layer and a pooling layer Deconv + BN + Relu, and generates 16 feature maps by using 32 convolutions;
the 9 th layer in the hidden layer adopts convolution, batch normalization, an activation layer and a pooling layer Deconv + BN + Relu, and 8 feature maps are generated by using 8 convolutions;
the 10 th layer in the hidden layer adopts convolution, batch normalization, an activation layer and a pooling layer Deconv + BN + ClippedRelu, and 1 convolution is used for generating 1 feature map;
the output layer comprises a regression layer;
the super pixel characteristics and the pixel characteristics are used as the input of a network, and the super pixel characteristics and the pixel characteristics are output as the group of the corresponding input image;
step 7, training the model;
step 8, outputting the trained network model M;
the detection phase comprises:
step 9, inputting an image sequence R2, and averaging 3-channel numerical values to perform image graying if the image is a color image to obtain a grayed image sequence G2; if the image is a gray image, directly enabling G2= R2;
step 10, performing median filtering on the image sequence to obtain a background image B2, and performing difference on the image sequence G2 and the background image B2 to obtain a candidate foreground sequence which is marked as a pixel characteristic F3;
step 11, performing superpixel segmentation on the color image sequence R2 to obtain region information C2;
step 12, calculating a histogram of pixels of a candidate foreground region corresponding to the superpixel according to the region information, wherein the interval is 0.1 when the histogram is in a range of [ -1,1 ];
step 13, taking the histogram of each region as the characteristics of all pixels in the region, and recording the histogram as a super-pixel characteristic F4;
step 14, taking the super pixel characteristics F4 and the pixel characteristics F3 as the input of the trained network model M;
and step 15, outputting a detection result.
The invention extracts candidate foreground (called as pixel characteristic) by using median filtering, carries out super-pixel segmentation on an image sequence, then extracts histogram characteristic (called as super-pixel characteristic) of super-pixels of the candidate foreground, and then respectively uses the pixel characteristic and the super-pixel characteristic as input of a pixel-finding fusion network.
In the whole detection process, only simple multiplication of a matrix is involved, so that the time complexity is small, the processing speed of a training stage and a detection stage is high, and the dynamic background can be effectively inhibited through the superpixel fusion characteristic due to the consideration of the dynamic characteristic.
Experiments show that the pixel fusion network has a good effect on 34 image sequences in CDNET 2014, more background noise can be removed through pixel fusion, and the pixel fusion network has stronger expression capability than a network with the same depth.
While the invention has been described in further detail with reference to specific embodiments thereof, it is not intended that the invention be limited to the specific embodiments thereof; for those skilled in the art to which the present invention pertains and related technologies, the extension, operation method and data replacement should fall within the protection scope of the present invention based on the technical solution of the present invention.
Claims (2)
1. A moving target detection method based on a super-pixel fusion network is characterized by comprising two stages: (1) a training stage; (2) a detection stage;
the detection phase comprises the following steps:
step 1, inputting a color image sequence R1, averaging 3-channel numerical values to perform image graying, and obtaining a grayed image sequence G1;
step 2, performing median filtering on the image sequence to obtain a background image B1, and performing difference on the image sequence G1 and the background image B1 to obtain a candidate foreground sequence which is marked as a pixel characteristic F1;
step 3, performing superpixel segmentation on the color image sequence R1 to obtain region information C1;
step 4, calculating a histogram of pixels of a candidate foreground region corresponding to the super-pixel according to the region information, wherein the interval is 0.1 when the histogram is in a range of [ -1,1 ];
step 5, taking the histogram of each region as the characteristics of all pixels in the region, and recording the histogram as a super-pixel characteristic F2;
step 6, constructing a network;
step 7, training the model;
step 8, outputting the trained network model M;
the detection phase comprises:
step 9, inputting an image sequence R2, and averaging 3-channel numerical values to perform image graying if the image is a color image to obtain a grayed image sequence G2; if the image is a gray image, directly enabling G2= R2;
step 10, performing median filtering on the image sequence to obtain a background image B2, and performing difference on the image sequence G2 and the background image B2 to obtain a candidate foreground sequence which is marked as a pixel characteristic F3;
step 11, performing superpixel segmentation on the color image sequence R2 to obtain region information C2;
step 12, calculating a histogram of pixels of a candidate foreground region corresponding to the super-pixel according to the region information, wherein the interval is 0.1 when the histogram is in a range of [ -1,1 ];
step 13, taking the histogram of each region as the characteristics of all pixels in the region, and recording the histogram as a super-pixel characteristic F4;
step 14, using the super pixel characteristics F4 and the pixel characteristics F3 as the input of the trained network model M;
and step 15, outputting a detection result.
2. The moving object detection method based on the super-pixel fusion network according to claim 1, wherein the specific method in step 6 is as follows:
(1) constructing an encoder:
the convolutional neural network comprises an input layer, a hidden layer and an output layer;
the input layer comprises two inputs, the resolution is 240 multiplied by 320, the number of channels of the encoder input corresponding to the super pixel characteristic F2 is 21, the encoder input corresponding to the pixel characteristic F1 is 1, and the convolution size in the convolution neural network is 3 multiplied by 3;
layer 1 of the hidden layers adopts convolution, batch normalization, an activation layer and a pooling layer Conv + BN + Relu + Maxpool, and 8 convolutions are used for generating 8 feature maps;
the layer 2 in the hidden layer adopts convolution, batch normalization, an activation layer and a pooling layer Conv + BN + Relu + Maxpool, and 16 convolutions are used for generating 16 feature maps;
layer 3 of the hidden layers adopts convolution, batch normalization, an activation layer and a pooling layer Conv + BN + Relu + Maxpool, and 32 convolutions are used for generating 32 feature maps;
layer 4 of the hidden layers adopts convolution, batch normalization, an activation layer and a pooling layer Conv + BN + Relu + Maxpool, and 64 feature maps are generated by using 64 convolutions;
(2) constructing a connecting layer:
the 5 th layer in the hidden layers is a connection layer, and the connection layer connects the two encoders by using localization;
(3) constructing a decoder:
layer 6 of the hidden layers adopts convolution, batch normalization, an activation layer and a pooling layer Conv + BN + Relu, and 128 convolutions are used for generating 64 feature maps;
the 7 th layer in the hidden layer adopts convolution, batch normalization, an activation layer and a pooling layer Deconv + BN + Relu, and generates 32 feature maps by using 64 convolutions;
the 8 th layer in the hidden layers adopts convolution, batch normalization, an activation layer and a pooling layer Deconv + BN + Relu, and 32 convolutions are used for generating 16 feature maps;
the 9 th layer in the hidden layers adopts convolution, batch normalization, an activation layer and a pooling layer Deconv + BN + Relu, and 8 convolutions are used for generating 8 feature maps;
the 10 th layer in the hidden layer adopts convolution, batch normalization, an activation layer and a pooling layer Deconv + BN + ClippedRelu, and 1 feature map is generated by using 1 convolution;
the output layer comprises a regression layer;
and taking the super pixel characteristics and the pixel characteristics as the input of the network, and outputting the super pixel characteristics and the pixel characteristics as the group of the corresponding input image.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210962818.6A CN115393585B (en) | 2022-08-11 | 2022-08-11 | Moving object detection method based on super-pixel fusion network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210962818.6A CN115393585B (en) | 2022-08-11 | 2022-08-11 | Moving object detection method based on super-pixel fusion network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115393585A true CN115393585A (en) | 2022-11-25 |
| CN115393585B CN115393585B (en) | 2023-05-12 |
Family
ID=84119085
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210962818.6A Active CN115393585B (en) | 2022-08-11 | 2022-08-11 | Moving object detection method based on super-pixel fusion network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115393585B (en) |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102637253A (en) * | 2011-12-30 | 2012-08-15 | 清华大学 | Video foreground object extracting method based on visual saliency and superpixel division |
| CN103578119A (en) * | 2013-10-31 | 2014-02-12 | 苏州大学 | Target detection method in Codebook dynamic scene based on superpixels |
| CN105809716A (en) * | 2016-03-07 | 2016-07-27 | 南京邮电大学 | Superpixel and three-dimensional self-organizing background subtraction algorithm-combined foreground extraction method |
| US20170140231A1 (en) * | 2015-11-13 | 2017-05-18 | Honda Motor Co., Ltd. | Method and system for moving object detection with single camera |
| CN111881915A (en) * | 2020-07-15 | 2020-11-03 | 武汉大学 | Satellite video target intelligent detection method based on multiple prior information constraints |
| CN112561949A (en) * | 2020-12-23 | 2021-03-26 | 江苏信息职业技术学院 | Fast moving target detection algorithm based on RPCA and support vector machine |
| CN112802054A (en) * | 2021-02-04 | 2021-05-14 | 重庆大学 | Mixed Gaussian model foreground detection method fusing image segmentation |
| CN112926466A (en) * | 2021-03-02 | 2021-06-08 | 江苏信息职业技术学院 | Moving target detection method based on differential convolutional neural network |
| CN114841941A (en) * | 2022-04-24 | 2022-08-02 | 江苏信息职业技术学院 | Moving target detection algorithm based on depth and color image fusion |
-
2022
- 2022-08-11 CN CN202210962818.6A patent/CN115393585B/en active Active
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102637253A (en) * | 2011-12-30 | 2012-08-15 | 清华大学 | Video foreground object extracting method based on visual saliency and superpixel division |
| CN103578119A (en) * | 2013-10-31 | 2014-02-12 | 苏州大学 | Target detection method in Codebook dynamic scene based on superpixels |
| US20170140231A1 (en) * | 2015-11-13 | 2017-05-18 | Honda Motor Co., Ltd. | Method and system for moving object detection with single camera |
| CN105809716A (en) * | 2016-03-07 | 2016-07-27 | 南京邮电大学 | Superpixel and three-dimensional self-organizing background subtraction algorithm-combined foreground extraction method |
| CN111881915A (en) * | 2020-07-15 | 2020-11-03 | 武汉大学 | Satellite video target intelligent detection method based on multiple prior information constraints |
| CN112561949A (en) * | 2020-12-23 | 2021-03-26 | 江苏信息职业技术学院 | Fast moving target detection algorithm based on RPCA and support vector machine |
| CN112802054A (en) * | 2021-02-04 | 2021-05-14 | 重庆大学 | Mixed Gaussian model foreground detection method fusing image segmentation |
| CN112926466A (en) * | 2021-03-02 | 2021-06-08 | 江苏信息职业技术学院 | Moving target detection method based on differential convolutional neural network |
| CN114841941A (en) * | 2022-04-24 | 2022-08-02 | 江苏信息职业技术学院 | Moving target detection algorithm based on depth and color image fusion |
Non-Patent Citations (5)
| Title |
|---|
| ZHI LIU 等: "Superpixel-Based Spatiotemporal Saliency Detection" * |
| 于洪洋 等: "基于超像素一致显著性的视频运动目标检测算法" * |
| 云红全 等: "基于超像素时空显著性的运动目标检测算法" * |
| 李阳: "基于RPCA和SVM的快速运动目标检测算法" * |
| 邢晴: "基于视觉显著性的海上目标检 测方法研究" * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115393585B (en) | 2023-05-12 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US20080112606A1 (en) | Method for moving cell detection from temporal image sequence model estimation | |
| US20100128789A1 (en) | Method and apparatus for processing video sequences | |
| CN110782477A (en) | Moving target rapid detection method based on sequence image and computer vision system | |
| CN111145209A (en) | Medical image segmentation method, device, equipment and storage medium | |
| US11042986B2 (en) | Method for thinning and connection in linear object extraction from an image | |
| CN112364865B (en) | Method for detecting small moving target in complex scene | |
| Xue et al. | Boundary-induced and scene-aggregated network for monocular depth prediction | |
| Boiangiu et al. | Voting-based image segmentation | |
| CN113395415A (en) | Camera data processing method and system based on noise reduction technology | |
| Katkar et al. | A novel approach for medical image segmentation using PCA and K-means clustering | |
| CN109785357B (en) | Robot intelligent panoramic photoelectric reconnaissance method suitable for battlefield environment | |
| CN111028263A (en) | Moving object segmentation method and system based on optical flow color clustering | |
| Zhang et al. | Local stereo matching: An adaptive weighted guided image filtering-based approach | |
| CN115830064B (en) | Weak and small target tracking method and device based on infrared pulse signals | |
| CN115393585A (en) | Moving target detection method based on super-pixel fusion network | |
| CN110880183A (en) | Image segmentation method, device and computer-readable storage medium | |
| CN111145121B (en) | Confidence term filter target tracking method for strengthening multi-feature fusion | |
| CN110490877B (en) | Target segmentation method for binocular stereo image based on Graph Cuts | |
| CN110942420B (en) | Method and device for eliminating image captions | |
| CN113989263A (en) | Image area saliency detection method based on super-pixel segmentation | |
| CN108510525B (en) | Template method for tracing, device, augmented reality system and storage medium | |
| Singh et al. | Performance Evaluation of the masking based Watershed Segmentation | |
| KR101711929B1 (en) | Method and apparatus for extraction of edge in image based on multi-color and multi-direction | |
| Tajudin et al. | Microbleeds detection using watershed-driven active contour | |
| CN111476821B (en) | Target tracking method based on online learning |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20231024 Address after: 215000 Rooms 7 # 211 and 213, No. 200 Shenhu Road, Suzhou Industrial Park, Suzhou Area, China (Jiangsu) Pilot Free Trade Zone, Wuxi City, Jiangsu Province Patentee after: Jiangsu botu Electrical Engineering Co.,Ltd. Address before: 214000 No.1 qianou Road, Wuxi City, Jiangsu Province Patentee before: JIANGSU VOCATIONAL College OF INFORMATION TECHNOLOGY |
|
| TR01 | Transfer of patent right |