Research On Computer Automated Recognition And Partitioning Technology Of Regional Magnetic Anomalies

Based on the characteristics of regional magnetic anomalies,the anomaly zoning is an important task to assist the division of geological units by using magnetic anomalies.So far,this work is basically based on artificial experience.Therefore,the inferred interpretation results are extremely dependent on the professional perspective of the analysts.The analysis results of different people may vary greatly.The use of computer graphics and image analysis technology to achieve automatic identification and partitioning of magnetic anomalies is the future direction,but due to its complexity and professionalism,there are many factors to consider,and there are huge difficulties.There are very few.This paper attempts to study and analyze the related results in image processing,and combine it with the changing characteristics of magnetic anomalies to form an anomaly automatic identification partitioning technique that can be applied to regional magnetic anomaly partitioning.For the specific examples and models,the paper first introduces the situation of the regional magnetic anomaly human eye partition and its basis,and then studies the partitioning method based on watershed segmentation and spectrum analysis,and illustrates the limitations and shortcomings of these methods.The core work is to use the information entropy-based superpixel segmentation algorithm.On the basis of this algorithm,the edge weight term is improved.The entropy rate is used to complete the accuracy of the intra-region compactness and homogeneity,and the balance term is used to control the partition standard.The computer automatically partitions the magnetic anomaly image and has achieved good results.After the initial partitioning,the optimal number of partitions was determined by the idea of Kappa coefficient.Then,the segmented area was subjected to morphological processing to remove the burrs and fill the cracks,making it more acceptable to the human eye.The feasibility and accuracy of the method are verified by comparing the results of computer partitioning with the results of expert partitioning.Combined with the geological structure information such as faults,the partitioning results are assisted and optimized.Finally,the local area was extracted for research,and the reasons for the difference between human eye discrimination and machine partition were analyzed.Through the research work in this paper,the gap in the field of automatic computer partitioning of magnetic anomalies is filled,especially in some details,the results of computer partitioning may be better than the division of human eyes.It provides an automated quantitative analysis method for magnetic anomaly partitioning,which reduces thepressure of manual partitioning and has a good development prospect.