object detection using non redundant local binary patterns

8/10/2019 Object detection using non redundant local binary patterns

1/19

A SEMINAR ON

OBJECT DETECTION USING NON-REDUNDANTLOCAL BINARY PATTERNS


2/19

INDEX

Introduction

Objectives

Methodology

LBP

Object detection using NRLBP

Object detection

Experimental results

Advantages and applications

Conclusion


3/19

Object detection using non-redundant local binary patterns

INTRODUCTION

In general object detection methods can be categorized as either global or local

approach.

Global approaches focus on detection of a full object in which the object is often

encoded by a feature vector. Some image features such as HOG (Histogram of

Oriented Gradients) and Haar-like features have been used widely to describe the

appearance of the objects.

Local approaches on the other hand detect objects by locating parts constituting

the objects .


4/19


Recently, some methods have employed interest points detectors, e.g. SIFT , to

identify the locations of the objectsparts. Local appearances of these parts then

were used as codewords to represent the object.

Motivated by the advantages of using a codebook of local appearance for object

detection and the discriminative power of local patterns in object recognition,

this paper presents an object detection method with the following contributions.


5/19


OBJECTIVES

The proposed descriptor is employed to encode humans appearance in a human detection

task.

Experimental results show that the NRLBP is robust and adaptive with changes of the

background and foreground and also outperforms the original LBP in detection task.


6/19

Assume that we have a number of images containing objects of interest.

We further assume that the object masks are also given.

The interest points can be detected using SIFT detector .

A set of interest points close to the object masks is then selected.

This set is called positive interest points set.

Other interest points not belonging to this set are called negative interest points.


METHODOLOGY


7/19

INPUT IMAGE LBP IMAGE

LBP

The local binary pattern method is widely used in 2D texture analysis.

The LBP was introduced by Ojala et al.

LBP is defined as an ordered set of binary comparison of pixel intensity

b/w centre pixel and its 8 surrounding pixels.

The decimal value of the resulting 8-bit word (LBP code) leads to 2^8

possible combination which are called LBP.



8/19


The original LBP was developed for texture classification and the success has

been due to its robustness under illumination changes, computational simplicityand discriminative power.

Fig. 1 represents an example of the original LBP in which the LBP code of the

center pixel (in red color and value 20) is obtained by comparing its intensity

with neighboringpixelsintensities.

Fig. 1. An illustration of the original LBPdescriptor.


9/19


The neighbor pixels whose intensities are equal or higher than the center pixels arelabeled as 1; otherwise as 0.

PROPERTIES OF LBP

Uniform and non-uniform LBP.

Uniform LBP is defined as the LBP that has at most two bitwise transitions from 0 to 1 and

vice versa in its circular binary representation. LBPs which are not uniform are called non-

uniform LBPs

LBP histogram.

Scanning a given image in pixel by pixel fashion, LBP codes are accumulated into adiscrete histogram called LBP histogram.


10/19

Non-Redundant LBP

Although the robustness of the original LBP has been demonstrated in manyapplications, it has certain drawbacks when employed to encode objects appearance.

Two notable disadvantages are:


Discriminative ability:the original LBP is sensitive to the relative changes between

the background and foreground (the region inside the object).

Storage requirement: the original LBP requires 2Pbins of histogram.


11/19

To overcome both of the above issues, we propose a novel LBP, namely,

Non-Redundant LBP (NRLBP), as follows,

NRLBPP,R(xc, yc)= min {LBPP,R(xc, yc), 2P 1 LBPP,R(xc, yc)}


Fig2 The left and right image represents the same humanwith inverted changes of the background and foreground


12/19

TO DETECT AN OBJECT USING NRLBP

First, we identify the locations of the objects parts using interest points detectorsand recognize the object by matching objects local appearances with predefined

code words.

Second, to describe local appearance, we introduce a new image descriptor namely,

Non-Redundant Local Binary Pattern (NRLBP). This descriptor is based on the

original LBP descriptor proposed for texture classification.

The proposed descriptor was evaluated by employing it to detect humans from static

images. Experimental results show that the NRLBP is robust and adaptive to

changes of the background and foreground and has a superior performance in

comparison with the original LBP.



13/19

OBJECT DETECTION

The object detection method is based on the scanning-window approach.

In other words we scan the input image by a detection window W at various

scales and positions.


The traditional way to observe the places or to track the places is the CCTV

cameras. The proposed system aims at detecting the abandoned objects, by using

a moving camera to collect the target.


14/19

EXPERIMENTAL RESULTS

The proposed method was evaluated by employing it to detect humans from still images



15/19

ADVANTAGES

The NRLBP outperforms the original LBP with regard to accuracy.

NRLBP provides more compact description of objects appearance.

It is more discriminative since it reflects the relative contrast betweenthe background and foreground.

NRLBP is robust and adaptive with changes of the background and

foreground and also outperforms the original LBP in detection task.


Obj d i i d d l l bi


16/19

APPLICATION

Object detection is used in biometrics, often as a part of (or together with)

a facial recognition system. It is also used in video surveillance, human

computer interface and image database management.


Obj d i i d d l l bi


17/19

CONCLUSIONS

This paper presents an object detection method using Non-Redundant LocalBinary Patterns (NRLBPs) to encode objectsappearance.

We evaluated and compared the performance of the proposed descriptor

with the original LBP and other descriptors for detecting humans from stillimages.



18/19

[1] N. Dalal and B. Triggs, Histograms of oriented gradients for human

detection, inProc IEEE Int. Conf. on Computer Visionand Pattern

Recognition, 2005, vol. 1, pp. 886893.

[2] P. Viola and M. J. Jones, Rapid object detection using a boosted cascade

of simple features, inProc IEEE Int. Conf.on Computer Vision and PatternRecognition, 2001, vol. 1, pp. 511518.

[3] P. Viola, M. J. Jones, and D. Snow, Detecting pedestrians using patterns of

motion and appearance,Int. Journal of ComputerVision, vol. 63, no. 2, pp.153161, 2005

REFERENCES


19/19

THANKING YOU

object detection using non redundant local binary patterns

Documents