中国科学院机构知识库网格系统: 检索

Robust correlation filter tracking with deep semantic supervision 期刊论文 OAI收割
IET IMAGE PROCESSING, 2019, 卷号: 13, 期号: 5, 页码: 754-760
作者: Wang, Wei; Chen, Zhaoming; Douadji, Lyes; Shi, Mingquan

|

收藏 | 浏览/下载：93/0 | 提交时间：2019/06/24

particle filtering (numerical methods) learning (artificial intelligence) target tracking convolutional neural nets robust correlation filter tracking high tracking performance tracking failure deep semantic supervision tracking framework redetection tracking mechanism particle filtering resampling CF tracker deep convolutional neural network tracking frames target occlusion handcrafted features real-time performance OTB2013 benchmark datasets OTB2015 benchmark datasets

Robust scale adaptive tracking by combining correlation filters with sequential Monte Carlo 期刊论文 OAI收割
SENSORS, 2017, 卷号: 17, 期号: 3, 页码: 1-16
作者: Ma JK(马俊凯); Luo HB(罗海波); Hui B(惠斌); Chang Z(常铮)

|

收藏 | 浏览/下载：17/0 | 提交时间：2017/03/26

Target Tracking Sequential Monte Carlo Framework Correlation Filter Scale Estimation Occlusion

复杂场景多目标跟踪中的遮挡算法与应用研究 学位论文 OAI收割
工学博士, 中国科学院自动化研究所: 中国科学院研究生院, 2012
作者: 丁欢

收藏 | 浏览/下载：116/0 | 提交时间：2015/09/02

复杂场景多目标跟踪遮挡分割粒子滤波嵌入式 complex scenes multi-target tracking occlusion segmentation particle filter embedded system

A fast target recognition algorithm based on MSA and MSR (EI CONFERENCE) 会议论文 OAI收割
2012 International Conference on Industrial Control and Electronics Engineering, ICICEE 2012, August 23, 2012 - August 25, 2012, Xi'an, China
作者: Wang Y.; Liu G.; Wang Y.; Wang Y.; Wang Y.

收藏 | 浏览/下载：36/0 | 提交时间：2013/03/25

This paper presents a new fast target recognition algorithm the proposed method is based on Multi-scale Auto convolution(MSA) and Multi-scale Retinex(MSR). As shown by the comparison with original MSA it appears that this new technique solves the problem that MSA algorithm is sensitive to illumination and the computational load is significantly reduced to 1/8th of that of the original MSA algorithm it is also robust to affine transform light projective transform noise thin fog occlusion and illumination change. the performed experiments show that it has fast searching speed and can accurately recognize and locate target in real scenes. 2012 IEEE.

The new approach for infrared target tracking based on the particle filter algorithm (EI CONFERENCE) 会议论文 OAI收割
International Symposium on Photoelectronic Detection and Imaging 2011: Advances in Infrared Imaging and Applications, May 24, 2011 - May 24, 2011, Beijing, China
作者: Sun H.; Han H.-X.; Sun H.

收藏 | 浏览/下载：64/0 | 提交时间：2013/03/25

Target tracking on the complex background in the infrared image sequence is hot research field. It provides the important basis in some fields such as video monitoring precision and video compression human-computer interaction. As a typical algorithms in the target tracking framework based on filtering and data connection the particle filter with non-parameter estimation characteristic have ability to deal with nonlinear and non-Gaussian problems so it were widely used. There are various forms of density in the particle filter algorithm to make it valid when target occlusion occurred or recover tracking back from failure in track procedure but in order to capture the change of the state space it need a certain amount of particles to ensure samples is enough and this number will increase in accompany with dimension and increase exponentially this led to the increased amount of calculation is presented. In this paper particle filter algorithm and the Mean shift will be combined. Aiming at deficiencies of the classic mean shift Tracking algorithm easily trapped into local minima and Unable to get global optimal under the complex background. From these two perspectives that "adaptive multiple information fusion" and "with particle filter framework combining" we expand the classic Mean Shift tracking framework.Based on the previous perspective we proposed an improved Mean Shift infrared target tracking algorithm based on multiple information fusion. In the analysis of the infrared characteristics of target basis Algorithm firstly extracted target gray and edge character and Proposed to guide the above two characteristics by the moving of the target information thus we can get new sports guide grayscale characteristics and motion guide border feature. Then proposes a new adaptive fusion mechanism used these two new information adaptive to integrate into the Mean Shift tracking framework. Finally we designed a kind of automatic target model updating strategy to further improve tracking performance. Experimental results show that this algorithm can compensate shortcoming of the particle filter has too much computation and can effectively overcome the fault that mean shift is easy to fall into local extreme value instead of global maximum value.Last because of the gray and fusion target motion information this approach also inhibit interference from the background ultimately improve the stability and the real-time of the target track. 2011 Copyright Society of Photo-Optical Instrumentation Engineers (SPIE).

An automatic pedestrian detection and tracking method: Based on mach and particle filter (EI CONFERENCE) 会议论文 OAI收割
2011 International Conference on Network Computing and Information Security, NCIS 2011, May 14, 2011 - May 15, 2011, Guilin, Guangxi, China
Han Q.; Yao Z.

收藏 | 浏览/下载：23/0 | 提交时间：2013/03/25

This paper introduces a pedestrian detecting and tracking approach. Correlation filters present the composite properties which have been successively used in target detection. Particle filter are combined to locate the targets in real-time. Our contribution is proposing a general algorithm that is able to detect and track pedestrians in clutter environments. We also create a different view pedestrian dataset. Experiments show our algorithm is comparative when there is block and occlusion in tracking. 2011 IEEE.

Study particle filter tracking and detection algorithms based on DSP signal processors (EI CONFERENCE) 会议论文 OAI收割
2010 International Conference on Computer, Mechatronics, Control and Electronic Engineering, CMCE 2010, August 24, 2010 - August 26, 2010, Changchun, China
Dong Y.; Chuan W.

收藏 | 浏览/下载：25/0 | 提交时间：2013/03/25

In Video tracking detection and tracking usually need two algorithms. The process is complex and need much time which detection and tracking are. In this paper a hybrid valued sequential state vector is formulated. The state vector is characterized by information of target appearance flag and of location. Particle filter-based method implements detection and tracking at one time. In order to reduce process time and think of pixel position in tracking field feature histogram of luminance is as observe vector and used posterior estimate. In this paper the luminance component is derived and target is recognized and tracked through image processor based on DSP in order to implementing real-time. The experimental results confirm that method can detect and track the object in real-time successfully when the number of particles is 160. The method is robust for rolling scale and partial occlusion. 2010 IEEE.

Study on color image tracking and detection algorithms based on particle filter (EI CONFERENCE) 会议论文 OAI收割
International Symposium on Photoelectronic Detection and Imaging 2009: Advances in Infrared Imaging and Applications, June 17, 2009 - June 19, 2009, Beijing, China
Wu C.; Sun H.-J.; Yang D.

收藏 | 浏览/下载：24/0 | 提交时间：2013/03/25

In Video tracking detection and tracking need two algorithms. The process is complex and need much time which detection and tracking is. In this paper a hybrid valued sequential state vector is formulated. The state vector is characterized by information of target appearance and of location. Particle filter-based method implements detection and tracking. In order to reduce process time and think of pixel position in tracking field feature histogram of color-based is as observe vector and used posterior estimate. The experimental results confirm that method can detect and track object in 17.68ms successfully when the number of particles is 160. The method is robust for rolling scale and partial occlusion. 2009 SPIE.

Partial occlusion detection of object boundary (EI CONFERENCE) 会议论文 OAI收割
2009 IEEE Intrumentation and Measurement Technology Conference, I2MTC 2009, May 5, 2009 - May 7, 2009, Singapore, Singapore
作者: Zhang J.; Zhang K.; Zhang K.; Zhang K.; Zhang K.

收藏 | 浏览/下载：39/0 | 提交时间：2013/03/25

Partial occlusion is a difficult problem in computer vision since whether the object is changed or occluded is ambiguous especially when distinguishing it only from the object boundary. In this paper we proposed a novel idea to solve this problem by taking shape matching as a morphing processing. A mass-spring model is constructed from the point set which is sampled from a template (or reference) object boundary by moving it to a target object which is deformed and/or occluded. From the morphing processing sufficient information can be obtained and an accurate detection of occlusion is performed. By using of the proposed method the application scope of occlusion detection is expanded while other method cannot be performed which need color texture or motion information. The experiments performed on synthetic and real world images proved the satisfactory performance of the proposed method. 2009 IEEE.

Real time tracking by LOPF algorithm with mixture model (EI CONFERENCE) 会议论文 OAI收割
MIPPR 2007: Automatic Target Recognition and Image Analysis; and Multispectral Image Acquisition, November 15, 2007 - November 17, 2007, Wuhan, China
Meng B.; Zhu M.; Han G.; Wu Z.

收藏 | 浏览/下载：33/0 | 提交时间：2013/03/25

A new particle filter-the Local Optimum Particle Filter (LOPF) algorithm is presented for tracking object accurately and steadily in visual sequences in real time which is a challenge task in computer vision field. In order to using the particles efficiently we first use Sobel algorithm to extract the profile of the object. Then we employ a new Local Optimum algorithm to auto-initialize some certain number of particles from these edge points as centre of the particles. The main advantage we do this in stead of selecting particles randomly in conventional particle filter is that we can pay more attentions on these more important optimum candidates and reduce the unnecessary calculation on those negligible ones in addition we can overcome the conventional degeneracy phenomenon in a way and decrease the computational costs. Otherwise the threshold is a key factor that affecting the results very much. So here we adapt an adaptive threshold choosing method to get the optimal Sobel result. The dissimilarities between the target model and the target candidates are expressed by a metric derived from the Bhattacharyya coefficient. Here we use both the counter cue to select the particles and the color cur to describe the targets as the mixture target model. The effectiveness of our scheme is demonstrated by real visual tracking experiments. Results from simulations and experiments with real video data show the improved performance of the proposed algorithm when compared with that of the standard particle filter. The superior performance is evident when the target encountering the occlusion in real video where the standard particle filter usually fails.