Digital Algorithm Research Of Precipitation Self-Recording Paper

The precipitation self-recording paper is one of the most valuable meteorological observation data, which record objectively the process of precipitation, and they are irreplaceable treasure for studying recent climate and environment to our country, and they are initial value for studying the change of regional climate and global climate. This paper is aimed at resolving some digital problems about precipitation self-recording paper, so I bring up the precipitation curve tracking algorithm based on adaptive direction filter. At the same time, this digital algorithm provides a template for other meteorological data, such as wind speed, barometric pressure, temperature and humidity self-recording paper.The main steps of digital algorithm for precipitation self-recording paper are image rotation correction, coordinate line extraction, tracing curve and extraction. First of all, we could find easily something, the axis of paper is composed by both horizontal and vertical lines, and the horizontal lines are longer than vertical lines. Canny operator detect the edge of precipitation self-recording paper, and get the angle of image rotation by Radon Transform, and then make a rotation, it can correct the inclined image. Secondly, in accordance with the rule of the lines, it takes advance of projection analysis of image to find the correct location of the axis. Thirdly, this paper brings up the precipitation curve tracking algorithm based on adaptive direction filter. The image of precipitation self-recording paper is segmented by adaptive threshold algorithm, then, eliminated the interference connected component by morphological filtering. After the binary image enhancement by direction filter algorithm, and we can extract the center line of the precipitation curve according to the maximum vertical position. Finally, the precipitation data interpretation by the method of four off six into five rounded into double. This method could eliminate ink contamination, curve fracture, pencil scratch and other interference factors, and then we get the real data of one minute, ten minutes and one hour from the image of precipitation self-recording paper.In addition to studying the algorithm of detecting precipitation self-recording paper, this paper also use MATLAB computer language to practice the correction of the algorithm this paper claims, such as image rotation and collection, the extraction about horizontal and vertical lines, the fitting extraction and quantitative analysis on precipitation self-recording paper. At the same time, I improve the proposed algorithm. The result from simulation can prove that this algorithm is effective to detect the self-recording paper and get precipitation data correctly.