Study On The Concrete Bridge Bottom Crack Inspection Method Based On Image Processing

In order to improve the bridge detection efficiency and reduce workers'risk under bridges, a bridge inspection robot was to be designed to identify cracks on the bottom surface of concrete bridges and to calculate their sizes automatically. In this paper, we summarized characteristics of bridge cracks, researched on the method of identifying crack using image processing technique, proposed and realized a method of crack identification, classification and connection, developed the software to analyze crack images, extracted cracks and calculated their sizes.Due to the complex environment and light condition under bridges, image collection was affected seriously. In addition to crack, disrupting chemicals, such as oil, ink and adhesive materials were also contained in images. As a result of similarity existing between objects and noises, they might be mistaken to each other. Therefore, the identification result was influenced seriously with low accuracy and impractical data.To solve those problems, we analyzed crack images collected from bridges bottom surface, described crack images'characteristics, carried out the pre-processing of original image, including graying, image enhancement, segment, filter and post-processing such as feature extraction, object identification, classification, measurement and fragments'connection. This paper focused on the analysis of crack shape and grayscale feature. As a result, several indexes were extracted to distinguish cracks from noises, and also projection feature were used to classify the detected cracks. According to the method proposed in this paper, the software was developed to identify and measure cracks by means of MATLAB. The experimental results showed that the method was effective under certain conditions.