수학적 형태론을 이용한 위성영상에서의 도로망 추출 = A study on the road network extraction from satellite image using mathematical morphology

다국어 초록 (Multilingual Abstract)

Because of the cost of the geographic information database creation, data acquisition should unquestionably be automated as much as possible. But the extraction of road network of urban area from Satellite Image is a quite complex. The present study considers a method for the extraction of Road Network using Mathematical Morphology method. SPOT HRV Panchromatic data of a subscene of Pusan Metropolitan city was used for the analysis. Top-hat transformation, thresholding, thinning, skeletonzation were used.
The main findings are summarized as follows :
1. The shape of road network was accurately extracted. But discontinuous linear features were also appeared.
2. The spatial resolution of SPOT HRV Panchromatic image is not sufficient for the urban road network extraction.
3. The difference of the results by thinning algorithm is not great. But preprocessing or postprocessing make it greater.
4. The noise reduction is not crucial for the road network extraction of Pusan Metropolitan city.
5. Some road type is not extracted. Vegetation make it difficult to use mathematical morphology for the road extraction. The material and width of the road is also important.
6. The extraction of the cross sections of the streets is very difficult. Manual digitization will be more efficient for this.

목차 (Table of Contents)

• 《목차》
• 1. 서론
• 1) 연구목적
• 2) 연구방법 및 자료
• 2. 수학적 형태론의 기본 정의
• 3. 수학적 형태론을 이용한 도로망 추출
• 1) 전처리(pre-processing)
• 2) 톱-햇 변환(top-hat transformation)
• 3) 임계치 선정 및 세선화
• 4. 사례 연구
• 5. 결론
• 〈참고문헌〉
• 〈Summary〉