+ All Categories
Home > Documents > Coarse Classification via Discrete Cosine Transform and Quantization

Coarse Classification via Discrete Cosine Transform and Quantization

Date post: 17-Jan-2016
Category:
Upload: earl
View: 33 times
Download: 0 times
Share this document with a friend
Description:
Coarse Classification via Discrete Cosine Transform and Quantization. Presented by Te-Wei Chiang July 18, 2005. Outline. Introduction Feature Extraction and Quantization Statistical Coarse Classification Experimental Results Conclusion. 1 Introduction. - PowerPoint PPT Presentation
Popular Tags:
25
1 Coarse Classification via Discrete Cosine Transform and Quantization Presented by Te-Wei Chiang July 18, 2005
Transcript
Page 1: Coarse Classification  via Discrete Cosine Transform and Quantization

1

Coarse Classification via Discrete Cosine

Transform and Quantization

Presented byTe-Wei

Chiang

July 18, 2005

Page 2: Coarse Classification  via Discrete Cosine Transform and Quantization

2

Outline

Introduction Feature Extraction and

Quantization Statistical Coarse Classification Experimental Results Conclusion

Page 3: Coarse Classification  via Discrete Cosine Transform and Quantization

3

1 Introduction

Paper documents -> Computer codes

OCR(Optical Character Recognition)

Page 4: Coarse Classification  via Discrete Cosine Transform and Quantization

4

Requirement of Coarse Classification

high accuracy rate, high reduction capacity, and quick classification time.

Page 5: Coarse Classification  via Discrete Cosine Transform and Quantization

5

Two subproblems in the design of classification systems

Feature extraction Classification

Page 6: Coarse Classification  via Discrete Cosine Transform and Quantization

6

Two Philosophies of Classification

Statistical Structural

Page 7: Coarse Classification  via Discrete Cosine Transform and Quantization

7

Advantages of the Statistical Approach

fixed-length vector, robustness for noisy patterns, less ambiguities, and easy to implement.

Page 8: Coarse Classification  via Discrete Cosine Transform and Quantization

8

Weaknesses of the Statistical Approach

Smear or average the parameters High dimensionality

Page 9: Coarse Classification  via Discrete Cosine Transform and Quantization

9

Goal of this Paper

The purpose of this paper is to design a general coarse classification scheme: low dependent on domain-specific

knowledge. To achieve this goal, we need:

reliable and general features, and general classification method

Page 10: Coarse Classification  via Discrete Cosine Transform and Quantization

10

2 Feature Extraction and Quantization

Find general features that can be applied properly to most application areas, rather than the best features.

Discrete Cosine Transform (DCT) is applied to extract statistical features.

Page 11: Coarse Classification  via Discrete Cosine Transform and Quantization

11

2.1. Discrete Cosine Transform (DCT)

The DCT coefficients F(u, v) of an N×N image represented by x(i, j) can be defined as

where

1

0

1

0

),()()(2

),(N

i

N

j

jixvuN

vuF ),2

)12(cos()

2

)12(cos(

N

vj

N

ui

.1

,021

)(otherwise

wforw

Page 12: Coarse Classification  via Discrete Cosine Transform and Quantization

12

The DCT coefficients of the character image of “ 佛” .

Page 13: Coarse Classification  via Discrete Cosine Transform and Quantization

13

2.2. Quantization

The 2-D DCT coefficient F(u,v) is quantized to F’(u,v) according to the following equation:

Thus, dimension of the feature vector can be reduced after quantization.

Page 14: Coarse Classification  via Discrete Cosine Transform and Quantization

14

Illustration of the Quantization Method

Page 15: Coarse Classification  via Discrete Cosine Transform and Quantization

15

3 Statistical Coarse Classification

The statistical classification system is operated in two modes: Training(learning) Classification(testing)

Page 16: Coarse Classification  via Discrete Cosine Transform and Quantization

16

3.1.Proposed coarse classification

scheme

Page 17: Coarse Classification  via Discrete Cosine Transform and Quantization

17

Illustration of Extracting the 2-D DCT Coefficients

Page 18: Coarse Classification  via Discrete Cosine Transform and Quantization

18

3.2. Grid code transformation (GCT)

Obtain the quantized DCT coefficient qij

Transform qij to positive integer dij

Such that object Oi can be transformed to a D-digit GC.

Page 19: Coarse Classification  via Discrete Cosine Transform and Quantization

19

3.3.Grid code sorting and elimination

Remove the redundant information.

A reduced set of candidate classes can be retrieved from the lookup table according to the GC of the test sample.

Page 20: Coarse Classification  via Discrete Cosine Transform and Quantization

20

4 Experimental Results

In our application, the objects to be classified are handwritten characters in Chinese paleography.

Since most of the characters in these rare books were contaminated by various noises, it is a challenge to achieve a high recognition rate.

Page 21: Coarse Classification  via Discrete Cosine Transform and Quantization

21

18600 samples (about 640 classes) are extracted from Kin-Guan ( 金剛 ) bible. Each character image was

transformed into a 48×48 bitmap. 1000 of the 186000 samples are used

for testing and the others are used for training.

Page 22: Coarse Classification  via Discrete Cosine Transform and Quantization

22

Page 23: Coarse Classification  via Discrete Cosine Transform and Quantization

23

5 Conclusions

This paper presents a coarse classification scheme based on DCT and quantization.

Due to the energy compacting property of DCT, the most significant features of a pattern can be extracted and quantized for the generation of the grid codes.

Page 24: Coarse Classification  via Discrete Cosine Transform and Quantization

24

Hence the potential candidates which are similar to the test pattern can be efficiently found by searching the training patterns whose grid codes are similar to that of the test pattern.

Page 25: Coarse Classification  via Discrete Cosine Transform and Quantization

25

Future works

Since features of different types complement one another in classification performance, by using features of different types simultaneously, classification accuracy could be further improved.


Recommended