Research On Image Analysis Of The Infrared Image Of Petroleum Pipeline Underground

Because of man induced or natural factors, the leakage of petroleum pipeline causes more than 100 million yuan loss every year. Many methods have been explored for the detection of the pipeline leakage. However, each method has some disadvantages and neither of them can detect the leakage of nonmetallic pipeline. Therefore, taking advantage of the infrared thermal radiation of the petroleum pipeline, we used the infrared thermal image method to detect the petroleum pipeline underground and to locate and direct the pipeline and its branch so as to provide a helpful assistant to detect the leakage of the pipeline. According to the characters of the pipeline infrared image, this paper studied the pre-process, enhancement, segmentation, mathematical morphology and skeleton track in image process. Firstly, we analyzed the source and character of the infrared image noise and designed methods to remove it. Besides analyzing and comparing the characters of different image enhancement methods, we also contrasted several adaptive segmentation methods and proved by the theoretic analysis and experiment results that iterative segmentation based on the rule of maximal distance in class could segment the pipeline infrared image effectively. Based on binary image, a set of mathematical morphology methods were used to get the main body of the pipeline. Then we used Freeman chain-code to track the skeleton of the pipeline so as to mark the pipeline and its branch. We collected a lot of the pipeline infrared images under different environment and condition. By analyzing of these images, it was proved that the image enchancement, mathematical morphology and skeleton track could improve the image quality effectively and locate the pipeline and its branch exactly.