Automatic Recognition Method Research Of Geophysical Logging Curve

As the oil and gas geophysical exploration will produce a large number of graphical results which saved in photo media, for further processing and research, they need to be digitized. Digital tables is one important means to the vector processing on logging engineering drawings, however, the rate of digital table is slow, the efficiency is also slow, the accuracy of extracted by operators is quite different, difficult to control, and eventually get a digital results can not immediately use, it must also be through multi-channel post-processing sectors in order to generate vector-based files. Thus it need for a higher degree of automation curve extraction application software, which not only has high accuracy, easy operation and good artificial interface, but also could be synchronized input the necessary of curve parameter information, and in most cases,there are no mutual interference with the data files.For the above issues, the automatic recognition and extraction of log curve has been researched. There is the main content:1.According to the characteristics of well logs, the background grids is eliminated by analyzing frequency-domain;2. According to the characteristics of the solid line which in the log curve image, the solid line is tracked and extracted by using the line's direction.(Including determine the image area and the starting point, the track on the curve, the key point recognizing, as well as tracking bug correction);3.The extraction and vectorization method on the dotted line in logging curve is designed, The method traces the curve by the dotted line's local slope,using tracing direction and step length, determining the trend of dotted line, skiping the gap in dotted line, and then carry out the vectorization on the dotted line;4. Based on the above approach, a full well-logging automatic identification system is finished. The system can implement the digital of logging curves, with certain characteristics of human-computer interaction.The research results of this paper for logging curves provides a new train of thought. By compairing with the past curve tracing algorithm, and actually running analysis, it make clear that this method improve the automatic recognition rate of the curve, and receive better results, thereby it demonstrate the feasibility of this approach and superiority.