An Application of Image Component in Crack Detection – A Case of Flood- Gate in Yilan

• Main Authors: Chang,Chia-Hao, 張家豪
• Other Authors: Chen,Dar-Der Michael
• Format: Others
• Language: zh-TW
• Published: 2016
• Online Access: http://ndltd.ncl.edu.tw/handle/ghgbwp

碩士 === 國立宜蘭大學 === 電機工程學系碩士班 === 104 === As technology advances, digital image processing techniques have been applied to many different fields, such as medicine, transportation and identification systems at current status, especially after the advent of smart phones, smart phone has become a photographic tool. In recent years, frequent earthquakes around the world resulted in damage and many significant cracks in buildings near the epicenters.In public construction, the government put considerable time and manpower in crack detection and crack security every year. However, at the time of the field survey, often due to terrain, limiting the reachable range, so cameras can not only break through the obstacles due to terrain, but can surveyors take the pictures of concrete cracks from a safer place, and leave the image formation for domestic workers to do image collection and analysis. Cracks detection in general image processing, often starts from a gray scale image, this approach can gain the advantages of time saving and efficiency. However, cracks can’t be detected in the context of a complex environment with much noise by this method. Taking into account photographs shot during on-site investigation, these images almost include complex background of environment. This paper proposes the use of color image components to keep the outline of a flood gate and separate it from the background of environment, and use proposed method to detect cracks in buildings. In this thesis, we took eight concrete structures as examples to look into: flood gate of Er Long upstream, Jhu Lu, Chai Wei, Ci Wu Lan, Sha Zai port, Shih Jian, Jhu An second and Da Keng Gu in Yilan, Taiwan. Images with cracks of color components and background are separated by simple algorithms, and then through the elimination of the border to get the outline of buildings. Using edge detection, mathematical morphology closing process and candidate cracks area calculation, etc., to detect cracks.