| Summary: | 碩士 === 國立中興大學 === 資訊科學與工程學系所 === 100 === The application of high resolution remote sensing images becomes more and more popular with the advancing of the remote sensing technology. Also, the processing and analysis of the remote sensing images become important issues. Remote sensing images are the fusion data obtained by remote devices such as the satellite or the aircraft from the surface of the target. These images are used in analysis, classification, and evaluation after image processing.
The first step of the remote sensing image processing is the detection of the shadow. Shadows are formed by buildings, rocks, trees, etc. and they may affected by the angle of the sun. They have different geometric shapes and colors, and they will cause a certain degree of error to the follow-up analysis. Even when the remote sensing images are used as base maps for just the road centerline determination, shadows of houses on both sides of road may easily deviate the results.
Otsu''s method is a common method of thresholding. It is also used in the shadow detection. The method proposed in this thesis was based on Otsu''s method and some enhancement was proposed. The proposed method enhanced the detection of the shadows in the image when light changed and reduced the probability of identifying of the non-shadow green plants as shadows without using infrared spectral data. The proposed method had been tested on images and the results were compared to the Otsu''s method. The statistical analysis of the experimental results and the judgment of the human vision show that the proposed method is very competitive with the Otsu''s method.
|
|---|
