使用颜色纹理直方图来进行目标跟踪_目标跟踪可以应用在哪些方面资源-CSDN文库

共13个文件

m：8个

avi：2个

asv：1个

目标跟踪

颜色纹理

4星 · 超过85%的资源需积分: 9 101 浏览量 2012-11-22 19:34:46 上传评论 1 收藏 28.81MB RAR 举报

资源推荐

资源详情

资源评论

收起资源包目录

Robust-Object-Tracking-demo.rar （13个子文件）

IJPRAI_demo_software

rgbPDF.m 1KB

LBPTexture.asv 3KB

escalator3.AVI 75.02MB

LBPTexture.m 3KB

Demo.m 4KB

normWindow.m 408B

flbp81_rgb_PDF.m 2KB

tabletennis.avi 14.08MB

flbp81_rgb_Tracking.m 4KB

Thumbs.db 7KB

select.m 656B

Robust Object Tracking using Joint Color-Texture Histogram.pdf 1.63MB

rgbTracking.m 3KB

November 4, 2009 14:55 WSPC/115-IJPRAI SPI-J068 00762

International Journal of Pattern Recognition

and Artiﬁcial Intelligence

Vol. 23, No. 7 (2009) 1245–1263

 World Scientiﬁc Publishing Company

ROBUST OBJECT TRACKING USING JOINT

COLOR-TEXTURE HISTOGRAM

∗

JIFENG NING

†

College of Information Engineering

Northwest A&F University

Yangling, Shaanxi, P. R. China

ning@sina.com

LEI ZHANG

‡

and DAVID ZHANG

‡

Biometrics Research Center

Department of Computing

Hong Kong Polytechnic University

Kowloon, Hong Kong, P. R. China

‡

cslzhang@comp.polyu.edu.hk

csdzhang@comp.polyu.edu.hk

CHENGKE WU

†

State Key Laboratory of Integrated Service Networks

Xidian University, Xi’an, P. R. China

ckwu@xidian.edu.cn

A novel object tracking algorithm is presented in this paper by using the joint color-

texture histogram to represent a target and then applying it to the mean shift frame-

work. Apart from the conventional color histogram features, the texture features of

the object are also extracted by using the local binary pattern (LBP) technique to

represent the object. The major uniform LBP patterns are exploited to form a mask

for joint color-texture feature selection. Compared with the traditional color histogram

based algorithms that use the whole target region for tracking, the proposed algorithm

extracts eﬀectively the edge and corner features in the target region, which characterize

better and represent more robustly the target. The experimental results validate that

the proposed method improves greatly the tracking accuracy and eﬃciency with fewer

mean shift iterations than standard mean shift tracking. It can robustly track the target

under complex scenes, such as similar target and background appearance, on which the

traditional color based schemes may fail to track.

Keywords: Object tracking; mean shift; local binary pattern; color histogram.

∗

This work was supported by the Hong Kong Polytechnic University Internal Grant (A-SA08),

and the National Science Foundation Council of China under Grants 60532060 and 60775020

‡

Author for correspondence

1245

November 4, 2009 14:55 WSPC/115-IJPRAI SPI-J068 00762

1246 J. Ning et al.

1. Introduction

Real-time object tracking is a critical task in computer vision applications. Many

tracking algorithms

have been proposed to overcome the diﬃculties arising from

noise, occlusion, clutter and changes in the foreground object or in the background

environment. Among the various tracking algorithms,

mean shift tracking algo-

rithms have recently become popular due to their simplicity and eﬃciency.

3,7,10,13,25

The mean shift algorithm was originally proposed by Fukunaga and Hostetler

for data clustering. It was later introduced into the image processing community by

Cheng.

Bradski

modiﬁed it and developed the Continuously Adaptive Mean Shift

(CAMSHIFT) algorithm to track a moving face. Comaniciu and Meer successfully

applied mean shift algorithm to image segmentation

and object tracking.

Mean

Shift is an iterative kernel-based deterministic procedure which converges to a local

maximum of the measurement function with certain assumptions on the kernel

behaviors. Furthermore, mean shift is a low complexity algorithm, which provides

a general and reliable solution to object tracking and is independent of the target

representation.

Currently, a widely used form of target representation is the color histogram,

7,14

which could be viewed as the discrete probability density function (PDF) of the

target region. Color histogram is an estimating mode of point sample distribution

and is very robust in representing the object appearance. However, using only color

histograms in mean shift tracking has some problems.

First, the spatial informa-

tion of the target is lost. Second, when the target has similar appearance to the

background, color histogram will become invalid to distinguish them. For a better

target representation, the gradient or edge features have been used in combination

with color histogram.

7,10

Several object representations that exploit the spatial

information have been developed by partitioning the tracking region into ﬁxed size

fragments,

meaningful patches

or the articulations of human objects.

For each

subregion, a color or edge feature based target model was presented.

The texture patterns,

9,18,22–24,28

which reﬂect the spatial structure of the object,

are eﬀective features to represent and recognize targets. Since the texture features

introduce new information that the color histogram does not convey, using the joint

color-texture histogram for target representation is more reliable than using only

color histogram in tracking complex scenes. The idea of combining color and edge for

target representation has been exploited by researchers.

7,10

However, how to utilize

eﬀectively both the color intensity and texture features is still a diﬃcult problem.

This is because though many texture analysis methods, such as gray concurrence

matrices

and Gabor ﬁltering,

have been proposed, they have high computational

complexity and cannot be directly used together with color histogram.

The local binary pattern (LBP)

16,17

technique is very eﬀective to describe the

image texture features. LBP has advantages such as fast computation and rota-

tion invariance, which facilitates the wide usage in the ﬁelds of texture analysis,

image retrieval, face recognition, image segmentation, etc.

2,15,19,21,27,28

Recently,

November 4, 2009 14:55 WSPC/115-IJPRAI SPI-J068 00762

Robust Object Tracking Using Joint Color-Texture Histogram 1247

LBP was successfully applied to the detection of moving objects via background

subtraction.

In LBP, each pixel is assigned a texture value, which can be naturally

combined with the color value of the pixel to represent targets. In Ref. 13, Nguyen

et al. employed the image intensity and the LBP feature to construct a two-

dimensional histogram representation of the target for tracking thermographic and

monochromatic video.

In this paper, we adopt the LBP scheme to represent the target texture feature

and then propose a joint color-texture histogram method for a more distinctive and

eﬀective target representation. The major uniform LBP patterns are used to iden-

tify the key points in the target region and then form a mask for joint color-texture

feature selection. The proposed target representation scheme eliminates smooth

background and reduces noise in the tracking process. Compared with the tradi-

tional RGB color space based target representation, it eﬃciently exploits the target

structural information and hence achieves better tracking performance with fewer

mean shift iterations and higher robustness to various interferences of background

and noise in complex scenes.

The paper is organized as follows. Section 2 brieﬂy introduces the mean shift

algorithm. Section 3 analyzes LBP and presents the joint color-texture histogram

scheme in detail. Experimental results are presented and discussed in Sec. 4.

Section 5 concludes the paper.

2. Mean Shift Algorithm

2.1. Target representation

A target is usually deﬁned by a rectangle or an ellipsoidal region in the image. Most

existing target tracking schemes use the color histogram to represent the rectangle

or ellipsoidal target. In this paper, we will present a new target representation

approach by using the joint color-texture histogram. First let us review the target

representation in the mean shift tracking algorithm.

Denote by {x

∗

}

i=1···n

the normalized pixel positions in the target region, which

is supposed to be centered at the origin point. The target model ˆq corresponding

to the target region is computed as











ˆq = {ˆq

}

u=1···m

ˆq

= C



i=1

k(x

∗



)δ[b(x

∗

) − u]

(1)

where ˆq

represent the probabilities of feature u in target model ˆq, m is the num-

ber of feature spaces, δ is the Kronecker delta function, b(x

∗

) associates the pixel

∗

to the histogram bin, k(x) is an isotropic kernel proﬁle and constant C is a

normalization function deﬁned by

C =1





i=1

k(x

∗



)(2)

November 4, 2009 14:55 WSPC/115-IJPRAI SPI-J068 00762

1248 J. Ning et al.

Similarly, the target candidate model ˆp(y) corresponding to the candidate region is

given by











ˆp(y)={ˆp

(y)}

u=1···m

ˆp

(y)=C



i=1





y − x





δ[b(x

) − u]

(3)





i=1





y − x





(4)

where ˆp

(y) represents the probability of feature u in the candidate model ˆp(y),

}

i=1···n

denote the pixel positions in the target candidate region centered at y,

h is the bandwidth and constant C

is a normalization function.

In order to calculate the likelihood of the target model and the candidate model,

a metric based on the Bhattacharyya coeﬃcient is deﬁned between the two normal-

ized histograms ˆp(y)andˆq as follows:

ρ[ˆp(y), ˆq]=



u=1



ˆp

(y)ˆq

(5)

The distance between ˆp(y)andˆq is then deﬁned as

d[ˆp(y), ˆq]=



1 − ρ[ˆp(y), ˆq](6)

2.2. Mean shift tracking

Minimizing the distance (6) is equivalent to maximizing the Bhattacharyya coeﬃ-

cient (5). The iterative optimization process is initialized with the target location

in the previous frame. Using Taylor expansion around ˆp

), the linear approx-

imation of the Bhattacharyya coeﬃcient (5) is obtained as

ρ[ˆp(y), ˆq] ≈



u=1



ˆp

)ˆq



i=1





y − x





(7)

where



u=1



ˆq

ˆp

)

δ[b(x

) − u](8)

Since the ﬁrst term in (7) is independent of y, to minimize the distance in (6) is

to maximize the second term in (7). In the iterative process, the estimated target

moves from y to a new position y

, which is deﬁned as



i=1





y−x







i=1





y−x





(9)

November 4, 2009 14:55 WSPC/115-IJPRAI SPI-J068 00762

Robust Object Tracking Using Joint Color-Texture Histogram 1249

When we choose kernel g with the Epanechnikov proﬁle,

(9) is reduced to



i=1



i=1

(10)

By using (10), the mean shift tracking algorithm ﬁnds in the new frame the most

similar region to the object. For more information about color histogram based

target representation and mean shift tracking, please refer to Ref. 7.

3. Target Tracking with Joint Color-Texture Histogram

3.1. Local binary pattern

The LBP

16,17

operator labels the pixel in an image by thresholding its neighbor-

hood with the center value and considering the result as a binary number (binary

pattern). The general version of the LBP operator is deﬁned as follows:

LBP

P,R

P −1



p=0

s(g

− g

(11)

where g

corresponds to the gray value of the center pixel (x

) of a local neigh-

borhood and g

to the gray values of P equally spaced pixels on a circle with radius

R. The function s(x)isdeﬁnedasfollows:

s(x)=



1 x ≥ 0

0 x<0

(12)

Figure 1 is an example of LBP

8,1

(P =8,R =1).ByvaryingP and R,we

have the LBP operators under diﬀerent quantization of the angular space and spa-

tial resolution, and multiresolution analysis can be accomplished by using multiple

LBP

P,R

operators.

The texture model derived by (11) has only gray-scale invariance. The gray-

scale and rotation invariant LBP texture model is obtained by (refer to Ref. 16 for

Fig. 1. An example of LBP

8,1

texture model.

评论收藏

内容反馈

zhongrui1984

2013-05-19

不错，是matlab编写的程序，可以用
wangwenhao1407

2017-04-18

谢谢共享！
shiguodong06

2012-12-23

连解压都解压不了这种垃圾还拿出来骗人真受不了了
helenlvli

2014-03-16

还好，只是要改点。因为我用的是2012matlab

大师鲁

粉丝: 104
资源: 10

使用颜色纹理直方图来进行目标跟踪

基于直方图匹配的目标跟踪

一种改进的基于颜色直方图的实时目标跟踪算法

基于颜色直方图的粒子滤波目标跟踪（hsv通道）

opencv特征提取 颜色直方图、纹理

纹理基元直方图matlab 版

基于梯度方向直方图特征的多核跟踪

OpenCV特征提取，包括边缘、颜色直方图，纹理【源码

基于颜色纹理直方图的meanshift目标跟踪算法 (2013年)

基于颜色属性直方图的尺度目标跟踪算法研究

运动目标跟踪中选择性颜色直方图的研究

基于改进颜色直方图映射的目标跟踪算法

基于颜色纹理特征的均值漂移目标跟踪算法 (2012年)

论文研究-一种基于空间直方图的多特征目标跟踪算法.pdf

基于背景加权的联合颜色纹理直方图跟踪算法

论文研究-基于统计模型的遗传粒子滤波器人体运动跟踪.pdf

论文研究-基于部分背景加权更新的均值漂移跟踪算法.pdf

计算机视觉作业：基于直方图的自适应阈值分割、利用聚类技术实现纹理图像分割、模板匹配技术、目标跟踪、背景建模、目标检测.zip

基于K均值的直方图，使用多分辨率特征向量进行颜色纹理数据库检索

计算机视觉作业基于直方图的自适应阈值分割、聚类实现纹理图像分割、模板匹配、目标跟踪、背景建模、目标检测python源码.zip

电信设备-融合颜色和纹理信息进行块背景建模的运动目标检测方法.zip

Medical-Image-Segmentation:使用实时脚本和应用程序在 MATLAB 中使用颜色和纹理信息进行图像分割。-matlab开发

基于复杂特征融合的改进mean shift目标跟踪

养殖场环境下的生猪多目标检测与跟踪方法

分块多特征目标描述子的移动机器人目标跟踪

opencv大作业

在 C++ 中 使用纹理和颜色特征进行图像分割_ C++_代码_下载

matlab提取轮廓代码-spb-mil:在多实例学习（MIL）框架中使用颜色，纹理和光谱线索进行对称检测

VehicleColorID：使用YOLO执行对象分割，并使用颜色直方图和kNN分类器进行颜色识别

最新资源

opencv特征提取颜色直方图、纹理

在 C++ 中使用纹理和颜色特征进行图像分割_ C++_代码_下载