Comparison of Template Matching and Mean Shift Algorithms with Kalman Filter for Object Tr, Slides of Applications of Computer Sciences

Download Comparison of Template Matching and Mean Shift Algorithms with Kalman Filter for Object Tr and more Slides Applications of Computer Sciences in PDF only on Docsity!

Presentation layout

 Introduction

 Scope

 Summary of previous work

 Comparison

 Kalman Filter

 Time Line

 Future work

 Conclusion

Introduction

 The basic concept behind object detection in videos engrosses:

 Verification of the presence of an object.  Locating it in particular for recognition.

 Object’s temporal and spatial changes are measured during a sequence of images by solving the temporal correspondence problem

Scope (Cont-I)

 It finds its scope in fields like :

 Vision-based control systems

 Human-computer interfaces

 Medical imaging

 Augmented reality

 Robotics

Objective

Image acquisition Preprocessing

Apply tracking Algorithms

Verification of results Obtained Post processing

Experimental results of Background

Subtraction

These figures show the results of background subtraction algorithm i.e. frame differencing applied on a sequence of images

Experimental results of Template Matching

Technique

Selected Template

Frame number 2,22,

Mean Shift Technique

 The Mean Shift algorithm is a non parametric technique used to locate density extrema or modes of a given distribution by an iterative procedure

 The Mean Shift algorithm is a mode-seeking process.

Fast Mean Shift

 The convergence speed of Fast Mean Shift is faster than that of Mean Shift.  Basic motivation comes from the idea that in the case of linear operations (e.g. ), any invertible linear operation can be applied to f or g if its inverse is applied to the result.  As in convolution, if derivative operator is applied both to the image and the kernel the result must then be double integrated its far more complex.

Comparison

Template Matching

• It requires input in the form of a template.
• Template selection is an important issue.
• It is simple to understand and implement.
• It uses cross correlations phenomena for comparison between template and current image.

Mean Shift

• It is a non parametric method.

• Window sizing is not trivial.

• It has complex mathematical issues and comparatively not that simple in application.
• It has different mathematics as fast mean shift is mainly based on convolutions.

Comparison(cont-I)

Template Matching

• If the orientation of the target changes then tracker starts loosing target results in futile tracking.
• Code length of template based technique is relatively

small.

Mean Shift

• It deals with occlusion to some extent.
• Code length of Mean shift is very large as compared to template based tracking.

Kalman Filter (Cont I)

 The solution provided by the Kalman filter is

recursive that each updated estimate of the state is

computed from the previous estimate and the new

input data

 Thus only previous estimate requires storage making

the Kalman filter computationally more efficient than

computing the estimate directly from the entire past

observed data at each step of the filtering process.

Why Kalman Filter?

 Kalman filter is computationally more multifarious still

there are reasons to use it.

 The computational complexity is nullified by recent

advancements in computer technology.

Reasons to use Kalman Filter (Cont-I)

 Optimal filtering

1. Kalman filter makes optimal use of the target measurements by adjusting the filter weights.
2. If the target measurement was more accurate, then weights will be automatically adjusted in such a way that more weight will be given to measurement than prediction.

Reasons to use Kalman Filter (Cont-II)

 Priori Information

1. Kalman filter optimally makes use of prior information.
2. This is especially useful when using two separate devices for searching and tracking.
3. Data from the searching devices can be optimally used to initialize the filter weights that will result in small transient in tracking filter.