An Effective Three-step Search Algorithm for Motion Estimation Supervisor ： Shung-chih Chen Presenter ： Chung-wei Tsai 1 Reference ： N. Sun, C. Fan, and.

An Effective Three-step Search Algorithm for Motion Estimation

Supervisor ： Shung-chih ChenPresenter ： Chung-wei Tsai

1

Reference：N. Sun, C. Fan, and X. Xia, "An effective three-step search algorithm for motion estimation," IT in Medicine & Education, 2009. ITIME '09. IEEE International Symposium, Vol. 1, pp. 400 - 403, Aug. 2009

Outline

Introduction

Review

Proposed Algorithm

Examples

Experimental Results

2

Outline

Introduction

Review

Proposed Algorithm

Examples

Experimental Results

3

Introduction (1/2)

This paper presents an effective three-step search algorithm (E3SS) in order to

reduce the computational complexity for motion estimation

improve the reliability of the image sequences for super-resolution reconstruction

4

Introduction (2/2)

Experimental results show that the new algorithm

not only ensures the search accuracy

but also reduces calculation time by a large margin

5

Block-matching Algorithm

Fig. 1. Block-matching algorithm6

Outline

Introduction

Review

Proposed Algorithm

Examples

Experimental Results

7

Three-step Search Algorithm (TSS)

Fig. 2. Three-step search8

The Drawbacks of TSS (1/2)

In the super-resolution reconstruction

There are small changes between each frame

The movement is mostly minuteness

9

The Drawbacks of TSS (2/2)

When the image movement is minuteness

TSS will create a bad impression on the effect and the accuracy of the motion estimation

According to this characteristic, this paper presents an effective three-step search algorithm (E3SS)

10

Outline

Introduction

Review

Proposed Algorithm

Examples

Experimental Results

11

The Search Template of E3SS

Fig. 3. The search template of E3SS12

Outline

Introduction

Review

Proposed Algorithm

Examples

Experimental Results

13

Example1 of E3SS

Fig. 4. Example1 of E3SS14

Example2 of E3SS

Fig. 5. Example2 of E3SS15

Outline

Introduction

Review

Proposed Algorithm

Examples

Experimental Results

16

Experimental Results

Table 1 Experimental data

17