CHAPTER 5 THEME UP. CHAPTER 5 THEME UP 5.1 What is theme ? 5.2 Why do we need to theme up? 5.3 How do we theme up? 5.4 Exercise.

CHAPTER 5

THEME UP

CHAPTER 5 THEME UP

5.1 What is theme ?

5.2 Why do we need to theme up?

5.3 How do we theme up?

5.4 Exercise

5.1 What is theme ? (1)

Theme is a focused point.

- Big impact

- Another problem that needs new data collecting for analysis.

- Unique point of view

- New evaluation result seems to be

investigated.

5.1 What is theme ? (2)

Example 5.1.1:

.sec3.0

.sec2.1.sec5.1

.sec3.0

.sec0.1det

.sec5.0

Decision

noise

cleanHMM

nrecognitio

extractionfeature

ectionedge

processpre

eprocesstim

5.1 What is theme ? (3)Data of recognition

0%

20%

40%

60%

80%

100%

Road Factory Classroom Studio

Clean

Noise

5.1 What is theme ? (4)

Example 5.1.1 (continue)

Noise and Clean can be focused as a theme and its data should be collected.

5.1 What is theme ? (5)

Example 5.1.2

The actual sales of two Sale men and a new sale man are shown in graph.What is appropriate to be theme?

5.1 What is theme ? (6)

Example 5.1.2 (continue)

0

10000

20000

30000

40000

50000

60000

1st Qtr 2nd Qtr 3rd Qtr 4th Qtr

Mr.A(new)

Mr.B

Mr.C

5.1 What is theme ? (7)

Example 5.1.2 (continue)

* New employee has improvement.

* Mr.B and C slump in 3rd Quarter. ---> Economic crisis?

* Why Mr.C has not come back?

5.1 What is theme ? (8)

Example 5.1.3 Breast calcification

- Survey:Data for finding patterns in frequency domain.

: Data in time domain.

- Problem:Noise and calcification is in high frequency range.

----> Detection of calcification

5.1 What is theme ? (9)

Example 5.1.4 Omnidirectional sensor:

Survey:

* for position detection in map.

* for detecting panorama by one camera.

5.1 What is theme ? (10)

Example 5.1.4 (Continue)

* detected panorama is not correctly projected

---> panorama calibration and usage

CHAPTER 5 THEME UP

5.1 What is theme ?

5.2 Why do we need to theme up?

5.3 How do we theme up?

5.4 Exercise

5.2 Why do we need to theme up?

* A way to know and solve the real problems.

* By focusing a theme and turning PDCA, once you can find the real problem.

* If not, you will solve the wrong or minor problems.

CHAPTER 5 THEME UP

5.1 What is theme ?

5.2 Why do we need to theme up?

5.3 How do we theme up?

5.4 Exercise

5.3 How do we theme up?(1)Step 1 raise up one of found problems.

Step 2 observe in time and place in that problem and collect the data.

Step 3 try to find the pattern of problem. If it is OK, go to create the idea.

Step 4 If there are something wrong in data, theme up another one and add new item to observe.

Step 5 go to step 2.

5.3 How do we theme up?(2)

Example 5.3.1 Data collecting results of improving efficiency of a machine

0%

20%

40%

60%

80%

100%

1st Qtr 2nd Qtr 3rd Qtr 4th Qtr

Target

Actual

Negative false

5.3 How do we theme up?(3)

Example 5.3.1(continue)

5.3 How do we theme up?(4)

Example 5.3.1(continue)

Themes can be considered as:

* According to the cause of 2nd quarter, for ex., machine trouble ----> maintenance.

* False negative has to be decreased

5.3 How do we theme up?(5)

Example 5.3.2 Durian Maturity Inspection

- Counting days as conventional method

- Human Knocking

- Automatic knocking

- By cavity analysis, --> vibration, ultrasonic, etc.

5.3 How do we theme up?(6)

Example 5.3.3 Final consonant Detection

- Picking up vowel detection as problem.

- Time and frequency approach

--> not constant range

- Energy approach---> highest frequency as vowel

- Finding last point of vowels.

5.3 How do we theme up?(7)

Other examples:

Mangosteen hardness detection

Durian ripeness classification

Human tracking

Corridor walking robot

Voice recognition etc.

CHAPTER 5 THEME UP

5.1 What is theme ?

5.2 Why do we need to theme up?

5.3 How do we theme up?

5.4 Exercise

5.4 Exercise