Automatic Recognition And Tracing Of Curves On Well Log Graphs

Oil and natural gas are both important energy and precious chemical resources, so it is not surprised that all the countries pay special attentions to their oil exploitation. A part of the key work of each oil field is to take full advantages of the wel 1 log curves on graphs, analyse and extract the information contained, and store them properly. The oil field, named Kelamayi, spent nearly three years to finish only one tenth of the work. The main aim was at vectorizing the well log curves applying foreign software. The main tool is a mouse. How to vectorize the information rapidly and correctly remains to be an emergent unresolved problem. To solve the problem, this paper tries to design software for automatic vectorization. What we have done are:· Deleting the back grids of the wel 1 log grapphs by analyzing in spatial domain or filtering in frenquency domain.· Recoginzing the curves rapidly and correctly by analysis of the colors, orientations and widths of the curves concerned.· Recogizing the dotted lines rapidly and correctly by analysis of the colors, areas, heights, widths, gaps and orientations of the lines· Displaying and processing of super large imges using memory mapping file.· Developing friendly oprating interface and easily used tools.· Designing the exportation and verification of the vectorzized data Among these lie the innovations as the follows.· Bring forward a method based on tracing, denoising and character parameters delecting to recognize the cuves rapidly and correctly. It has been proved to be more efficient than the foreign one.· Apply multi-threads in the course of recognition to ensure operating in real time.· Real i se the automatic connect ion of hi-scale curves and dotted 1 ines, which is unavailable in dat.a-dosk and forign software.· Display and process successfully super large images in a memory limited system and run smoothly on PC.