WO2018024030A1 - 一种夜视红外图像中的基于显著性的道路目标提取方法 - Google Patents
一种夜视红外图像中的基于显著性的道路目标提取方法 Download PDFInfo
- Publication number
- WO2018024030A1 WO2018024030A1 PCT/CN2017/087531 CN2017087531W WO2018024030A1 WO 2018024030 A1 WO2018024030 A1 WO 2018024030A1 CN 2017087531 W CN2017087531 W CN 2017087531W WO 2018024030 A1 WO2018024030 A1 WO 2018024030A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- saliency
- image
- night vision
- level model
- target
- 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.)
- Ceased
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/50—Context or environment of the image
- G06V20/56—Context or environment of the image exterior to a vehicle by using sensors mounted on the vehicle
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/50—Context or environment of the image
- G06V20/56—Context or environment of the image exterior to a vehicle by using sensors mounted on the vehicle
- G06V20/588—Recognition of the road, e.g. of lane markings; Recognition of the vehicle driving pattern in relation to the road
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; 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 OR CALCULATING; 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
- G06V10/42—Global feature extraction by analysis of the whole pattern, e.g. using frequency domain transformations or autocorrelation
- G06V10/431—Frequency domain transformation; Autocorrelation
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; 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
- G06V10/46—Descriptors for shape, contour or point-related descriptors, e.g. scale invariant feature transform [SIFT] or bags of words [BoW]; Salient regional features
- G06V10/462—Salient features, e.g. scale invariant feature transforms [SIFT]
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30248—Vehicle exterior or interior
- G06T2207/30252—Vehicle exterior; Vicinity of vehicle
- G06T2207/30256—Lane; Road marking
Definitions
- the invention belongs to the field of machine vision, and in particular relates to a saliency-based road target extraction method in a night vision infrared image.
- the visual saliency mechanism can be used to extract significant areas of the road scene that have a significant impact on driving safety, laying the foundation for post-processing of computer vision, improving the effectiveness and efficiency of computer vision subsequent processing, and facing complex
- the visual attention mechanism in the human visual system.
- Machine vision also first screens the original input data like the human visual system, so as to reduce the computational complexity and improve the processing efficiency for post-processing, so a visual saliency model of machine vision is proposed.
- Visual saliency is a concept of neurophysiology and psychology. Neurophysiologists and psychologists in various countries have proposed many models to explain the attention of the human visual system, such as the peripheral-central nerve suppression mechanism, to simulate human visual attention. Inspired by the ability of the mechanism to efficiently screen data, researchers in various countries have conducted extensive and in-depth research on how to screen machine vision for data such as human visual saliency.
- Chinese patent (CN102999926A) discloses an image visual saliency calculation method based on the underlying feature fusion. This method combines the underlying features such as color and texture to obtain a significant region, but only based on the underlying features, without considering high Hierarchical features, so the saliency extraction is inaccurate.
- Chinese patent (CN103020993A) discloses a two-channel color contrast fusion visual saliency detection method, which uses a superpixel method to calculate a color saliency value, thereby obtaining a saliency image based on color features, but only in the time domain. Processing within the range, and based only on color features, ignoring the significance in the frequency domain, so the extraction of significant regions is not comprehensive.
- the above method has the above-mentioned deficiencies in the process of extracting the saliency regions, and processes an original natural image without considering that the night vision infrared image corresponding to the original image can be processed, so the road scene at night
- the detection in the middle has low accuracy, is susceptible to interference from the light in the background environment, and the outline of the significant target is not clear enough.
- the present invention avoids the effects of light in a non-significant background environment in the original visible light image at night by processing the corresponding night vision infrared image of the original visible light image, thereby improving the accuracy of the extraction of significant objects in the night road scene.
- the present invention proposes a saliency-based road target extraction method in a night vision infrared image, which solves the problem that the existing visual attention model has a low correct rate in target detection in a night road scene. Due to the interference of non-significant areas in natural scenes, it is susceptible to the influence of lights in night road images, and the outline of significant targets is not clear.
- the present invention achieves the above technical objects by the following technical means.
- a saliency-based road target extraction method in a night vision infrared image comprising the following steps:
- S1 for the night vision infrared image of the road target, first use the GBVS model (image saliency analysis algorithm based on graph theory) to perform coarse extraction of the saliency region based on local contrast, and obtain the saliency image CC of the night vision infrared image;
- GBVS model image saliency analysis algorithm based on graph theory
- the eigenimage ZZ is fused with the global and local cues to obtain the saliency image AA.
- a saliency extraction method based on fusion global and local cues is applied to the saliency image.
- ZZ performs the fusion of global and local cues, and the specific steps for obtaining the saliency image AA are:
- the saliency image of the background of the high-level model of the saliency image ZZ and the saliency image of the saliency target of the high-level model are image-fused to obtain a saliency image of the high-level model of the saliency image ZZ;
- the saliency image of the low-level model of the saliency image ZZ is merged with the saliency image of the high-level model to obtain a final saliency image AA.
- the calculation formula of the super complex matrix f(x, y) in the S2.2 is Where vector They are the same unit vectors in the three-dimensional Cartesian coordinate system as the x, y, and z axes.
- denotes the modulus of the vector.
- the S 3.4 obtains the saliency image of the background of the high-level model of the saliency image ZZ, and the saliency image of the saliency target of the high-level model is calculated by the computer randomly selecting between the punitive consideration and the regularization Balance factor ⁇ , making it satisfy among them
- U i refers to the description of the night vision infrared image
- D is the encoding dictionary
- o is the element multiplication
- cc i refers to the covariance matrix
- ⁇ is the regularization coefficient
- ⁇ is the product
- tr(cc i ) Is the sum of the diagonal elements in the matrix cc i
- H n is the vector of the encoding dictionary
- n is the number of elements in the encoding dictionary
- ⁇ is the weighting coefficient of the local adapter
- dist(U i , H n ) refers to The Euclidean distance between U i and H n ; the saliency image of the
- the present invention first uses the GBVS model to extract the preliminary extraction saliency; and on the basis of this, the method based on the spectral scale space of the super complex frequency domain is used to extract the global features, which can extract significant in the frequency domain. Sexual characteristics combine the time domain with the saliency in the frequency domain. Finally, the saliency extraction method based on the fusion global and local cues is used to combine the global and local saliency features to clear the saliency target contour. Improve the accuracy of the salient target extraction and the clarity of the outline. The present invention effectively highlights the salient regions and suppresses the non-significant regions, thereby achieving a more accurate extraction of the salient goals.
- 1 is a flow chart of a saliency-based road target extraction method in a night vision infrared image
- FIG. 2 is a flow chart of a saliency extraction method based on fused global and local cues
- FIG. 3 is a saliency image obtained by the present invention
- FIG. 3(a) is a night vision infrared image of a pedestrian
- FIG. 3(b) is a pedestrian saliency image obtained by the present invention
- FIG. 3(c) is A night vision infrared image of the vehicle
- Fig. 3(d) is a vehicle saliency image obtained by the present invention.
- a flowchart of a saliency-based road target extraction method in a night vision infrared image includes the following steps:
- S1 for the night vision infrared image of the road target, first use the GBVS model (image saliency analysis algorithm based on graph theory) to perform coarse extraction of the saliency region based on local contrast, and obtain the saliency image CC of the night vision infrared image;
- GBVS model image saliency analysis algorithm based on graph theory
- Markov random field The saliency of a pixel in a small image area is only related to the saliency of its neighboring pixels, but not related to the saliency of pixels in other places. Then the set of pixels in this small area of the image is a Marco. With the airport;
- the saliency image CC is obtained by finding the equilibrium distribution of the Markov chain.
- r, g, b are the three components of the red, green and blue of the significant image CC;
- vectors are the same unit vectors in the three-dimensional Cartesian coordinate system as the x, y, and z axes;
- g represents a Gaussian kernel
- * represents a convolution
- the saliency map sequence ⁇ Z i ⁇ of the saliency image CC is calculated based on the spectral scale space ⁇ AA k ⁇ of the saliency image CC; the calculation formula for calculating the saliency map sequence ⁇ Z i ⁇ is:
- indicates the amplitude value
- F() represents the super complex Fourier transform
- g denotes the Gaussian kernel
- * denotes the convolution
- ⁇ denotes the product
- F -1 denotes the inverse Fourier transform
- v() denotes the vector portion after the Fourier transform
- denotes the modulus of the vector
- superpixel segmentation refers to segmenting a series of small image regions consisting of adjacent pixels with similar features such as color, brightness, texture, etc.
- the sex image ZZ is divided into a plurality of small image areas;
- OK(q) refers to the frequency at which the pixel q falls within the significant target region
- w i refers to a plurality of small image regions segmented in the super pixel. Refers to the number of pixels that fall in the area w i ;
- (x, y) refers to the average coordinate of the saliency image ZZ
- (x o , y o ) refers to the center coordinates of the image
- V x , V y refers to the width and height of the small image region.
- the saliency image P d (w i ) of the low-level model can be obtained from equations (10) and (11).
- c j refers to the small image area located on the boundary of the saliency image ZZ
- M refers to the number of small image areas located on the boundary of the saliency image ZZ
- dd(w i , c j ) refers to the application The difference between the small image regions w i and c j obtained by Euclidean metrics
- the equilibrium coefficient ⁇ between the punitive consideration and the regularization is randomly selected by the computer to satisfy
- U i refers to the description of the night vision infrared image
- D is the encoding dictionary
- o is the element multiplication
- cc i refers to the covariance matrix
- ⁇ is the regularization coefficient
- ⁇ is the product
- tr(cc i ) is the sum of the diagonal elements in the matrix cc i ;
- H n is the vector of the coding dictionary
- n is the number of elements in the coding dictionary
- ⁇ is the weight coefficient of the local adapter
- dist(U i , H n ) refers to the Euclidean distance between U i and H n ;
- the saliency image of the background of the high-level model of the saliency image ZZ and the saliency image of the saliency target of the high-level model are image-fused to obtain a saliency image of the high-level model of the saliency image ZZ;
- the saliency image of the low-level model of the saliency image ZZ is merged with the saliency image of the high-level model to obtain a final saliency image AA.
- the saliency image of the pedestrian and the vehicle as shown in FIG. 3 is obtained by the saliency-based road target extraction method in the night vision infrared image of the present invention.
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Multimedia (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Image Analysis (AREA)
- Studio Devices (AREA)
- Image Processing (AREA)
Abstract
Description
Claims (8)
- 一种夜视红外图像中的基于显著性的道路目标提取方法,其特征在于,包括如下步骤:S1,针对道路目标的夜视红外图像,先运用GBVS模型(基于图论的图像显著性分析算法)进行基于局部对比度的显著性区域的粗提取,得到夜视红外图像的显著性图像CC;S2,运用基于超复数频域的谱尺度空间的方法,基于全局特征,对显著性图像CC进行显著性区域的再提取,得到夜视红外图像的显著性图像ZZ;S3,运用基于融合全局和局部线索的显著性提取方法,对显著性图像ZZ进行全局和局部线索的融合,得到显著性图像AA。
- 根据权利要求1所述的一种夜视红外图像中的基于显著性的道路目标提取方法,其特征在于,所述S2中运用基于超复数频域的谱尺度空间的方法,基于全局特征,对显著性图像CC进行显著性区域的再提取的具体步骤为:S2.1,通过对显著性图像CC的红绿蓝三个分量组合计算,得到显著性图像CC的三个特征图U,RG,BY;S2.2,对特征图U,RG,BY进行整合,计算得到显著性图像CC的超复数矩阵f(x,y);S2.3,先对超复数矩阵f(x,y)进行超复数傅立叶变换,然后计算显著性图像CC的幅值谱A,相位谱p,特征轴谱kk;S2.4,对幅值谱A进行卷积,从而得到显著性图像CC的谱尺度空间{AAk};S2.5,基于显著性图像CC的谱尺度空间{AAk}计算得到显著性图像CC的显著图序列{Zi};S2.6,运用基于最大方差的原则选取显著图序列{Zi}中的一个显著图序列作为夜视红外图像的显著性图像ZZ。
- 根据权利要求1所述的一种夜视红外图像中的基于显著性的道路目标提取方法,其特征在于,所述S3中运用基于融合全局和局部线索的显著性提取方法,对显著性图像ZZ进行全局和局部线索的融合,得到显著性图像AA的具体步骤为:S3.1,运用超像素分割的方法,将显著性图像ZZ分割成多个小图像区域;S3.2,基于全局对比度对S3.1中的多个小图像区域进行计算,得到显著性图像ZZ的低层次模型的显著性图像;S3.3,按照阀值分割的方法对显著性图像ZZ的低层次模型的显著性图像进行处理,分别得到显著性图像ZZ的高层次模型的背景的编码字典BC、高层次模型的显著性目标的编码字典FC;S3.4,计算显著性图像ZZ的高层次模型的背景的显著性图像、高层次模型的显著性目标的显著性图像pg(wi);S3.5,将显著性图像ZZ的高层次模型的背景的显著性图像、高层次模型的显著性目标的显著性图像进行图像融合,得到显著性图像ZZ高层次模型的显著性图像;S3.6,将显著性图像ZZ的低层次模型的显著性图像与高层次模型的显著性图像相融合,得到最终的显著性图像AA。
- 根据权利要求2所述的一种夜视红外图像中的基于显著性的道路目标提取方法,其特征在于,所述S2.3中幅值谱A的计算公式为A=|F(f(x,y))|,其中| |表示求幅值,F()表示超复数傅立叶变换。
- 根据权利要求3所述的一种夜视红外图像中的基于显著性的道路目标提取方法,其特征在于,所述S3.3中高层次模型的背景的编码字典BC、高层次模型的显著性目标的编码字典FC的阀值分割原则为:BC=wi(if Pd(wi)<λ1),FC=wi(if Pd(wi)>λ2),其中λ1为背景的编码中选取的阀值,λ2为显著性目标的编码中选取的阀值。
- 根据权利要求3所述的一种夜视红外图像中的基于显著性的道路目标提取方法, 其特征在于,所述S3.4得到显著性图像ZZ的高层次模型的背景的显著性图像、高层次模型的显著性目标的显著性图像的计算过程为:由计算机随机选取惩罚性考虑与正则化之间的平衡系数μ,使其满足其中局部适配器Ui指的是对夜视红外图像的描述,D为编码字典,ο为元素相乘,cci指的是协方差矩阵,λ为正则化系数,×为乘积,tr(cci)指的是矩阵cci中对角线元素之和,Hn为编码字典的矢量,n为编码字典中的元素的个数,α为局部适配器的权重系数,dist(Ui,Hn)指的是Ui与Hn之间的欧式距离;则高层次模型的背景的显著性图像、高层次模型的显著性目标的显著性图像Pg(wi)=||Ui-Dbbi||2。
Priority Applications (5)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| DE112017000273.7T DE112017000273T9 (de) | 2016-08-03 | 2017-06-08 | Ausgeprägtheitsbasiertes Verfahren zum Extrahieren eines Straßenziels aus einem Nachtsicht-Infrarotbild |
| AU2017307115A AU2017307115B2 (en) | 2016-08-03 | 2017-06-08 | A method of extracting road targets based on saliency in a night vision infrared image |
| US16/070,078 US10635929B2 (en) | 2016-08-03 | 2017-06-08 | Saliency-based method for extracting road target from night vision infrared image |
| JP2019506074A JP2019523509A (ja) | 2016-08-03 | 2017-06-08 | 暗視赤外画像における顕著性に基づく道路オブジェクト抽出方法 |
| CA3032487A CA3032487C (en) | 2016-08-03 | 2017-06-08 | Saliency-based method for extracting road target from night vision infrared image |
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610631619.1A CN106295542A (zh) | 2016-08-03 | 2016-08-03 | 一种夜视红外图像中的基于显著性的道路目标提取方法 |
| CN201610631619.1 | 2016-08-03 |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| WO2018024030A1 true WO2018024030A1 (zh) | 2018-02-08 |
Family
ID=57664635
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| PCT/CN2017/087531 Ceased WO2018024030A1 (zh) | 2016-08-03 | 2017-06-08 | 一种夜视红外图像中的基于显著性的道路目标提取方法 |
Country Status (7)
| Country | Link |
|---|---|
| US (1) | US10635929B2 (zh) |
| JP (1) | JP2019523509A (zh) |
| CN (1) | CN106295542A (zh) |
| AU (1) | AU2017307115B2 (zh) |
| CA (1) | CA3032487C (zh) |
| DE (1) | DE112017000273T9 (zh) |
| WO (1) | WO2018024030A1 (zh) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111062905A (zh) * | 2019-12-17 | 2020-04-24 | 大连理工大学 | 一种基于显著图增强的红外和可见光融合方法 |
| CN115439497A (zh) * | 2022-08-31 | 2022-12-06 | 北京瀚景锦河科技有限公司 | 基于改进hou模型的红外图像舰船目标快速识别方法 |
Families Citing this family (40)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106295542A (zh) | 2016-08-03 | 2017-01-04 | 江苏大学 | 一种夜视红外图像中的基于显著性的道路目标提取方法 |
| CN107133558B (zh) * | 2017-03-13 | 2020-10-20 | 北京航空航天大学 | 一种基于概率传播的红外行人显著性检测方法 |
| CN108256519B (zh) * | 2017-12-13 | 2022-06-17 | 苏州长风航空电子有限公司 | 一种基于全局和局部相互作用的红外图像显著检测方法 |
| CN108090885B (zh) * | 2017-12-20 | 2021-08-27 | 百度在线网络技术(北京)有限公司 | 用于处理图像的方法和装置 |
| CN108805029B (zh) * | 2018-05-08 | 2021-08-24 | 天津师范大学 | 一种基于显著对偶激活编码的地基云图识别方法 |
| CN109346102B (zh) * | 2018-09-18 | 2022-05-06 | 腾讯音乐娱乐科技(深圳)有限公司 | 音频开头爆音的检测方法、装置及存储介质 |
| CN109934224B (zh) * | 2019-03-07 | 2023-06-09 | 绵阳慧视光电技术有限责任公司 | 基于马尔科夫随机场和视觉对比机制的小目标检测方法 |
| CN110717454B (zh) * | 2019-10-10 | 2022-03-11 | 北京北特圣迪科技发展有限公司 | 一种舞台环境下的轮式机器人障碍检测方法 |
| CN110782405B (zh) * | 2019-10-14 | 2022-10-18 | 中国科学院光电技术研究所 | 一种基于梯度辨识的点目标和暗斑图像背景均衡方法 |
| CN111046869B (zh) * | 2019-11-27 | 2023-09-29 | 北京腾信软创科技股份有限公司 | 一种基于深度学习的显著区域提取方法及系统 |
| CN111008585B (zh) * | 2019-11-29 | 2023-04-07 | 西安电子科技大学 | 基于自适应分层高分辨sar图像的舰船目标检测方法 |
| CN110889386A (zh) * | 2019-12-02 | 2020-03-17 | 山东浪潮人工智能研究院有限公司 | 一种基于结构特征学习的手指静脉识别方法 |
| CN112884790B (zh) * | 2019-12-10 | 2022-09-30 | 长江存储科技有限责任公司 | 一种图形分析方法、系统及存储介质 |
| CN110991547A (zh) * | 2019-12-12 | 2020-04-10 | 电子科技大学 | 一种基于多特征最优融合的图像显著性检测方法 |
| CN111145148A (zh) * | 2019-12-13 | 2020-05-12 | 中国人民解放军国防科技大学 | 基于压缩感知的图像受干扰程度评价方法 |
| CN111260624B (zh) * | 2020-01-15 | 2023-05-26 | 陕西师范大学 | 一种基于特征匹配和尺度选择策略的多尺度红外小目标的探测方法 |
| US12313427B2 (en) | 2020-03-10 | 2025-05-27 | Pioneer Corporation | Map data generation device |
| CN112164068B (zh) * | 2020-07-17 | 2023-07-28 | 中国人民解放军63892部队 | 一种警戒雷达ppi图像目标和干扰区域自适应提取方法 |
| CN112070098B (zh) * | 2020-08-20 | 2024-02-09 | 西安理工大学 | 基于频率调整模型的高光谱图像显著目标检测方法 |
| CN112132810B (zh) * | 2020-09-24 | 2023-09-12 | 西安电子科技大学 | 基于感知逻辑和特征对比的图像显著性检测方法 |
| CN112329796B (zh) * | 2020-11-12 | 2023-05-23 | 北京环境特性研究所 | 基于视觉显著性的红外成像卷云检测方法和装置 |
| CN112419265B (zh) * | 2020-11-23 | 2023-08-01 | 哈尔滨工程大学 | 一种基于人眼视觉机制的伪装评价方法 |
| CN112329887B (zh) * | 2020-11-26 | 2023-12-05 | 深圳大学 | 基于尺度自适应滤波的高光谱和LiDAR联合分类方法 |
| CN112541486B (zh) * | 2020-12-31 | 2022-11-08 | 洛阳伟信电子科技有限公司 | 基于改进Pixel分割的红外弱小目标检测算法 |
| CN112613528B (zh) * | 2020-12-31 | 2023-11-03 | 广东工业大学 | 一种基于显著性变分的点云精简方法、装置及存储介质 |
| CN113139425B (zh) * | 2021-03-09 | 2024-09-20 | 杭州电子科技大学 | 一种基于双视通路交互感知的轮廓检测方法 |
| CN112950477B (zh) * | 2021-03-15 | 2023-08-22 | 河南大学 | 一种基于双路径处理的高分辨率显著性目标检测方法 |
| CN113449667A (zh) * | 2021-07-08 | 2021-09-28 | 四川师范大学 | 基于全局卷积和边界细化的显著目标检测方法 |
| CN113449735B (zh) * | 2021-07-15 | 2023-10-31 | 北京科技大学 | 一种超像素分割的语义分割方法及装置 |
| CN113573058B (zh) * | 2021-09-23 | 2021-11-30 | 康达洲际医疗器械有限公司 | 一种基于时空显著性融合的帧间图像编码方法 |
| CN113935984B (zh) * | 2021-11-01 | 2025-06-20 | 中国电子科技集团公司第三十八研究所 | 多特征融合的复杂背景中红外弱小目标检测方法及系统 |
| CN114037820A (zh) * | 2021-11-05 | 2022-02-11 | 北京环境特性研究所 | 一种基于多特征融合的红外弱小多目标的检测方法和装置 |
| CN114549969B (zh) * | 2022-01-12 | 2024-07-26 | 中国地质大学(武汉) | 一种基于图像信息融合的显著性检测方法和系统 |
| CN114463365B (zh) * | 2022-04-12 | 2022-06-24 | 中国空气动力研究与发展中心计算空气动力研究所 | 红外弱小目标分割方法、设备及介质 |
| CN115424037A (zh) * | 2022-10-12 | 2022-12-02 | 武汉大学 | 一种基于多尺度稀疏表示的显著目标区域提取方法 |
| CN116128916B (zh) * | 2023-04-13 | 2023-06-27 | 中国科学院国家空间科学中心 | 一种基于空间能流对比度的红外弱小目标增强方法 |
| CN116630651B (zh) * | 2023-05-19 | 2024-09-10 | 中国人民解放军国防科技大学 | 一种显著性目标检测方法、系统、电子设备、存储介质 |
| CN116740165B (zh) * | 2023-06-21 | 2024-07-05 | 南宁桂电电子科技研究院有限公司 | 双尺度下自适应红外图像目标大小的检测方法 |
| CN116958876B (zh) * | 2023-07-28 | 2024-06-14 | 郑州轻工业大学 | 一种基于多光谱双目立体视觉的视频异常行为检测方法和系统 |
| US20250232561A1 (en) * | 2024-01-12 | 2025-07-17 | Microsoft Technology Licensing, Llc | Transformer for classification |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102521592A (zh) * | 2011-11-30 | 2012-06-27 | 苏州大学 | 基于非清晰区域抑制的多特征融合显著区域提取方法 |
| CN104463855A (zh) * | 2014-11-25 | 2015-03-25 | 武汉科技大学 | 一种基于频域和空域结合的显著区域检测方法 |
| CN104463907A (zh) * | 2014-11-13 | 2015-03-25 | 南京航空航天大学 | 一种基于视觉显著性特征的自适应目标跟踪方法 |
| US20160004929A1 (en) * | 2014-07-07 | 2016-01-07 | Geo Semiconductor Inc. | System and method for robust motion detection |
| CN106295542A (zh) * | 2016-08-03 | 2017-01-04 | 江苏大学 | 一种夜视红外图像中的基于显著性的道路目标提取方法 |
Family Cites Families (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2008053901A (ja) * | 2006-08-23 | 2008-03-06 | Konica Minolta Holdings Inc | 撮像装置および撮像方法 |
| WO2011001398A2 (en) * | 2009-06-30 | 2011-01-06 | Mango Dsp Inc. | Method circuit and system for matching an object or person present within two or more images |
| JP5630518B2 (ja) * | 2012-03-14 | 2014-11-26 | 株式会社デンソー | 運転支援装置 |
| US9025880B2 (en) * | 2012-08-29 | 2015-05-05 | Disney Enterprises, Inc. | Visual saliency estimation for images and video |
| CN102999926B (zh) | 2012-11-12 | 2016-06-29 | 北京交通大学 | 一种基于底层特征融合的图像视觉显著性计算方法 |
| CN103020993B (zh) | 2012-11-28 | 2015-06-17 | 杭州电子科技大学 | 一种双通道颜色对比度融合的视觉显著性检测方法 |
| CN103996040A (zh) * | 2014-05-13 | 2014-08-20 | 西北工业大学 | 融合局部-全局对比度的自底向上的视觉显著性生成方法 |
| CN105631898B (zh) * | 2015-12-28 | 2019-04-19 | 西北工业大学 | 基于空时显著性融合的红外运动目标检测方法 |
| CN105761269A (zh) * | 2016-03-14 | 2016-07-13 | 北京大学 | 基于多尺度区分性子空间的图像显著对象的检测方法 |
| US10198654B2 (en) * | 2016-06-24 | 2019-02-05 | Ebay Inc. | Automated saliency map estimation |
-
2016
- 2016-08-03 CN CN201610631619.1A patent/CN106295542A/zh active Pending
-
2017
- 2017-06-08 US US16/070,078 patent/US10635929B2/en active Active
- 2017-06-08 AU AU2017307115A patent/AU2017307115B2/en not_active Ceased
- 2017-06-08 DE DE112017000273.7T patent/DE112017000273T9/de active Active
- 2017-06-08 JP JP2019506074A patent/JP2019523509A/ja active Pending
- 2017-06-08 WO PCT/CN2017/087531 patent/WO2018024030A1/zh not_active Ceased
- 2017-06-08 CA CA3032487A patent/CA3032487C/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102521592A (zh) * | 2011-11-30 | 2012-06-27 | 苏州大学 | 基于非清晰区域抑制的多特征融合显著区域提取方法 |
| US20160004929A1 (en) * | 2014-07-07 | 2016-01-07 | Geo Semiconductor Inc. | System and method for robust motion detection |
| CN104463907A (zh) * | 2014-11-13 | 2015-03-25 | 南京航空航天大学 | 一种基于视觉显著性特征的自适应目标跟踪方法 |
| CN104463855A (zh) * | 2014-11-25 | 2015-03-25 | 武汉科技大学 | 一种基于频域和空域结合的显著区域检测方法 |
| CN106295542A (zh) * | 2016-08-03 | 2017-01-04 | 江苏大学 | 一种夜视红外图像中的基于显著性的道路目标提取方法 |
Non-Patent Citations (1)
| Title |
|---|
| JI, CHAO ET AL.,: "Image Salient Region Detection Based on Spatial and Frequency Domains", JOURNAL OF JILIN UNIVERSITY (ENGINEERING AND TECHNOLOGY EDITION), vol. 44, no. 1, 31 January 2014 (2014-01-31) * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111062905A (zh) * | 2019-12-17 | 2020-04-24 | 大连理工大学 | 一种基于显著图增强的红外和可见光融合方法 |
| CN111062905B (zh) * | 2019-12-17 | 2022-01-04 | 大连理工大学 | 一种基于显著图增强的红外和可见光融合方法 |
| CN115439497A (zh) * | 2022-08-31 | 2022-12-06 | 北京瀚景锦河科技有限公司 | 基于改进hou模型的红外图像舰船目标快速识别方法 |
Also Published As
| Publication number | Publication date |
|---|---|
| AU2017307115A1 (en) | 2019-02-21 |
| AU2017307115B2 (en) | 2020-05-21 |
| CN106295542A (zh) | 2017-01-04 |
| CA3032487A1 (en) | 2018-02-08 |
| US20190197344A1 (en) | 2019-06-27 |
| CA3032487C (en) | 2021-02-16 |
| DE112017000273T9 (de) | 2019-01-03 |
| JP2019523509A (ja) | 2019-08-22 |
| DE112017000273T5 (de) | 2018-09-13 |
| US10635929B2 (en) | 2020-04-28 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| WO2018024030A1 (zh) | 一种夜视红外图像中的基于显著性的道路目标提取方法 | |
| Ye et al. | Optical and SAR image fusion based on complementary feature decomposition and visual saliency features | |
| CN108549874B (zh) | 一种目标检测方法、设备及计算机可读存储介质 | |
| CN103971115B (zh) | 一种基于NDVI和PanTex指数的新增建设用地图斑自动提取方法 | |
| CN103077531B (zh) | 基于边缘信息的灰度目标自动跟踪方法 | |
| CN103218605B (zh) | 一种基于积分投影与边缘检测的快速人眼定位方法 | |
| CN103578083B (zh) | 基于联合均值漂移的单幅图像去雾方法 | |
| CN107680054A (zh) | 雾霾环境下多源图像融合方法 | |
| CN102842044B (zh) | 高分辨率可见光遥感图像变化检测方法 | |
| CN109460764B (zh) | 一种结合亮度特征与改进帧间差分法的卫星视频船舶监测方法 | |
| CN107818303A (zh) | 无人机油气管线影像自动对比分析方法、系统及软件存储器 | |
| CN111242864A (zh) | 一种基于Gabor纹理约束的手指静脉图像修复方法 | |
| CN107610164A (zh) | 一种基于多特征混合的高分四号影像配准方法 | |
| CN106971397B (zh) | 基于改进jseg算法的城市高分辨率遥感影像分割方法 | |
| Junwu et al. | An infrared and visible image fusion algorithm based on LSWT-NSST | |
| CN106529472B (zh) | 基于大尺度高分辨率高光谱图像的目标探测方法及装置 | |
| CN106875407A (zh) | 一种结合形态学和标记控制的无人机影像林冠分割方法 | |
| CN114820384A (zh) | 基于局部对比度和梯度的红外弱小目标检测方法 | |
| CN110310263B (zh) | 一种基于显著性分析和背景先验的sar图像居民区检测方法 | |
| Zhang et al. | Enhancing efficient global understanding network with CSWin transformer for urban scene images segmentation | |
| Lin et al. | Enhancing deep-learning object detection performance based on fusion of infrared and visible images in advanced driver assistance systems | |
| CN109102534B (zh) | 雾霾天气条件下的光学遥感图像配准方法及系统 | |
| CN103065298A (zh) | 一种模拟视网膜滤波的视觉显著性检测方法 | |
| CN109934103A (zh) | 基于暗通道先验和区域协方差检测图像中显著物体的方法 | |
| CN110796684B (zh) | 目标跟踪方法以及相关装置 |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| WWE | Wipo information: entry into national phase |
Ref document number: 112017000273 Country of ref document: DE |
|
| 121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 17836213 Country of ref document: EP Kind code of ref document: A1 |
|
| ENP | Entry into the national phase |
Ref document number: 3032487 Country of ref document: CA |
|
| ENP | Entry into the national phase |
Ref document number: 2019506074 Country of ref document: JP Kind code of ref document: A |
|
| ENP | Entry into the national phase |
Ref document number: 2017307115 Country of ref document: AU Date of ref document: 20170608 Kind code of ref document: A |
|
| 122 | Ep: pct application non-entry in european phase |
Ref document number: 17836213 Country of ref document: EP Kind code of ref document: A1 |


