基于矩阵分解的背景分离算法在油气井可视化检测中的应用

Application of Background Separation Algorithm Based on Matrix Factorization to Visual Inspection in Oil and Gas Wells

针对VideoLog可视化测井系统在油气井检测过程中遇到水中有较多漂浮物,影响对井壁的观测的问题,介绍了基于矩阵低秩稀疏分解的背景分离法,对VideoLog可视化测井系统录制的视频所组成的观测矩阵进行低秩稀疏分解,分离出的低秩矩阵和稀疏矩阵分别对应于视频背景和水中漂浮物。对背景图像序列分别进行中值、均值和众数处理,该方法去除了水中的漂浮物,获得质量较高的背景图像,有效解决油气井可视化检测过程中遇到水中漂浮物较多而影响观测的问题,为管柱问题检测的定量分析提供依据,增强VideoLog可视化检测的功能。

In view of the problem that there are floating objects in the water,which affects the observation with the VideoLog visual logging system for the detection of oil and gas wells.This article mainly introduces an effective method based on matrix low rank sparse.The decomposed background separation method performs low-rank sparse decomposition on the observation matrix composed of the video recorded by the VideoLog visualization logging system.The separated low-rank matrix and sparse matrix correspond to the video background and floating objects in water.By performing median,mean and mode processing on the background image sequence,this method removes floating objects in the water and obtains a high-quality background image,which can effectively solve the problem that the floating objects in the water during the visual detection of oil and gas wells affect the observation,and also provide quantitative analysis for the detection of pipe problems.Provides a foundation to further enhance the VideoLog visual detection function.