International Journal of Computer Applications (0975 – 8887)
Volume 93 – No 1, May 2014
37
Performance Analysis of Various Feature Detector and
Descriptor for Real-Time Video based Face Tracking
Akash Patel
P.G. student
Computer Engineering
SCET, Surat, India.
D. R. Kasat
Associate Professor
SCET college
Surat, India
Sanjeev Jain, Ph. D
Director
MITS
Gwalior, India
V. M. Thakare, Ph. D
Head CSE Dept.
Amravati University
Amravati, India
ABSTRACT
This paper presents the performance analysis of various
contemporary feature detector and descriptor pair for real time
face tracking. These feature detectors/descriptors are mostly
used in image matching applications. Some feature
detectors/descriptors like STAR, FAST, BRIEF, FREAK, and
ORB can also be used for SLAM applications due to their
high performance. However using only one of these feature
detectors for object tracking may not provide good accuracy
due to various challenges in tracking like abrupt change in
object motion, non-rigid object structure, change in
appearance of object, occlusions in the scene and camera
motion. But it can be combined other object tracking
algorithm to improve the overall tracking accuracy. In this
paper we have measured the tracking speed and accuracy of
these feature detectors in real time video for face tracking
using parameters like average number of detected key points,
average detection time of key-point, frame per second and
number of matches using OpenCV.
General Terms
Object tracking, Image matching.
Keywords
Face tracking, Feature detectors and Feature descriptors.
1. INTRODUCTION
Visual object tracking can be defined as the process of
tracking a moving object(s) continuously using a camera. The
goal of object tracking is to determine the position of the
object in frames continuously and reliably in video [1]. It is
very important task in many computer vision applications.
This process should keep track of its motion, orientation,
occlusion in scene etc. Tracking can be simplified by
imposing constraints on the motion and/or appearance of
objects [2].
Feature detectors are used to find interest points in given
image. It aims at computing abstractions of image information
whereas feature extraction aims at how to represents the
detected key points of image. Feature extraction is basically a
special form of dimensionality reduction. These
detectors/descriptors are used as first step in many
applications like object tracking, localization, image matching
and recognition.
The detection, description and matching of feature points
plays a vital role in most of the contemporary algorithms for
SLAM (Simultaneous Localization and Mapping) [3, 4]. In
past years several new detectors (FAST [6], SURF [7], and
CenSurE-based STAR [8]) and descriptors (SIFT [5], SURF
[7], BRIEF [9], ORB [10], BRISK [11], and FREAK [13])
have been proposed. They have been successfully applied to
the object detection and tracking task.
Currently, to the extent of our knowledge there is no
comparative study of the newest point detectors and
descriptors with regard to their applicability in face tracking.
In [14] author has compared various feature descriptors for
Pedestrian detection. In [15] and [16] the authors has
described the desired characteristics of the feature detectors
and descriptors for visual SLAM, but they have not presented
any experimental results.
This paper present the performance analysis of the detector
descriptor pairs in the context of face tracking. The measure
of the pair’s efficiency was based on the various parameters
like average number of detected key points, average detection
time of key-point, detection frame per second and number of
matches. The videos were taken from several real-time
situations using Webcam supporting resolution up to 720p and
speed up to 30 fps.
The following paper is organized as follows. The Section 2
presents the short summary of feature detector and descriptor
evaluated in the study. Section 3 presents the evaluation
methodology and result analysis and the section 4 contain the
concluding remarks.
2. VARIOUS FEATURE DETECTORS
AND DESCRIPTORS
2.1 FAST feature detector
The FAST [6] (Features from Accelerated Segment Test)
feature detector was the first algorithm based on AST
(Accelerated Segment Test). It first examines the values of the
intensity function of pixels in a circle of radius r around the
candidate point p. They have considered pixel on a circle
’bright’ if its intensity value is brighter by at least t(threshold),
and ’dark’ if its intensity value is darker by at least t than the
intensity value of p. They have classified a candidate pixel as
a feature on a basis of a segment test – if a contiguous, at least
n pixels long arc of ’bright’ or ’dark’ pixels is found in the
circle than it is considered as feature. They have used ID3
[17] algorithm to optimize the order in which pixels are
tested, resulting in high computational efficiency. The
segment test alone produces small sets of adjacent positive
responses. To further refine the results, they have used an
additional corner-ness measure for non-maximum suppression
(NMS). To improve the speed the NMS is applied only to a
small fraction of pixels that positively passed the segment test.
是否龙磊磊真的一无所有
- 粉丝: 422
- 资源: 32
我的内容管理
收起
- 我的资源
快来上传第一个资源
我的收益 登录查看自己的收益
我的积分
登录查看自己的积分
我的C币
登录后查看C币余额
我的收藏 
我的下载 
下载帮助
会员权益专享
最新资源
- stc12c5a60s2 例程
- Android通过全局变量传递数据
- c++校园超市商品信息管理系统课程设计说明书(含源代码) (2).pdf
- 建筑供配电系统相关课件.pptx
- 企业管理规章制度及管理模式.doc
- vb打开摄像头.doc
- 云计算-可信计算中认证协议改进方案.pdf
- [详细完整版]单片机编程4.ppt
- c语言常用算法.pdf
- c++经典程序代码大全.pdf
- 单片机数字时钟资料.doc
- 11项目管理前沿1.0.pptx
- 基于ssm的“魅力”繁峙宣传网站的设计与实现论文.doc
- 智慧交通综合解决方案.pptx
- 建筑防潮设计-PowerPointPresentati.pptx
- SPC统计过程控制程序.pptx
资源上传下载、课程学习等过程中有任何疑问或建议，欢迎提出宝贵意见哦~我们会及时处理！
点击此处反馈
评论0