Semantic Segmentation Of Ultrasonic Imaging Logging Fractures Based On Deep Learning

With the rapid development of social economy,the consumption of oil and gas resources is increasing day by day.The exploration and development of fractured reservoirs is of great significance for the stable supply of oil and gas resources.Because the fractures in fractured reservoirs can affect the distribution of oil and gas,the effective detection of fractures has a certain guiding role in the exploration and development of fractured reservoirs.At present,the identification of fractures by logging images is one of the main means of detecting fractures,but the identification of logging fractures still requires manual operation,which is not only time-consuming and labor-intensive,but also has certain human errors in the identification results.Based on this,this paper focuses on the semantic segmentation method of ultrasonic imaging logging fractures based on deep learning to realize the automatic detection of fractures in logging images.The main work and innovations of this paper are as follows :(1)Considering the high labeling cost of logging images,this paper designs a logging fracture segmentation algorithm without labeling.The algorithm can learn features similar to logging image data sets from existing data sets,thereby reducing the cost of logging image annotation.(2)Due to the small number of samples and unclear contour of logging image data sets,this paper designs a logging fracture segmentation algorithm based on UNet structure,which is widely used in the field of medical image segmentation,according to the similarity between medical image segmentation task and logging fracture segmentation task.(3)Because of the excellent performance of Transformer in computer vision tasks,this paper designs a logging fracture segmentation algorithm based on Transformer.Based on the UNet structure,the algorithm replaces the coding layer and decoding layer of UNet with the basic module of Swin Transformer.This algorithm also has good segmentation effect.The three algorithms designed in this paper can realize the automatic detection of logging fractures,but limited by the number of samples and the accuracy of labeling,the segmentation effect of logging fractures still has a large room for improvement.The work of this paper attempts to realize the automatic segmentation of logging fractures from three different ideas,which provides some reference for the follow-up work.