Core sequence image annotation method combining optical flow and watershed segmentation

Abstract

The invention discloses a core sequence image labeling method combining optical flow and watershed segmentation, and mainly relates to the labeling technology of core sequence images. It includes the following steps: (1) registering the core sequence images; (2) setting markers on the first frame, and performing marker-based watershed segmentation on the current image; (3) continuing if the segmented images meet the labeling requirements Perform step (4), otherwise adjust a small number of marked points, and segment again; (4) use the marked points of the current frame that satisfy the labeling as the feature points of the improved LK optical flow method based on pyramid layering to track, and get the next frame. Mark points; (5) Repeat (2) to (4) to finally obtain the labeling of each sequence image. The method of the invention significantly improves the labeling efficiency and improves the labeling quality by utilizing the characteristics of the core sequence diagram and the correlation between layers.

Description

technical field [0001] The invention relates to an image processing technology of core sequence images, in particular to a core sequence image labeling method combining optical flow and watershed segmentation. Background technique [0002] In the field of petroleum geology, in order to analyze the rock pore and particle structure, CT or FIB-SEM scans are usually used to obtain a 2D sequence image, and the sequence image is segmented to obtain the 3D pore structure of the core. In the segmentation-related algorithms, more and more deep learning algorithms are applied in this field. With the continuous application and development of deep learning in various fields, deep learning algorithms urgently need massive amounts of labeled data. For most artificial intelligence projects, manual labeling of large amounts of data is a very heavy task. Especially for the core sequence image, there are various components such as rock, matrix, clay minerals, etc., which makes it very time-c...

AI Technical Summary

Problems solved by technology

For (1), although this type of labeling software is convenient and fast to a certain extent, it ignores the interlayer correlation of sequence diagrams and can only label pictures one by one. Said, the workload is still too much

And for core images with irregular and uneven target edges, it is more difficult and time-consuming to use software to accurately outline the edges

For (2), due to the different evaluation and labeling of the labelers, the labeling ability is uneven, and it is difficult to guarantee the accuracy of the pictures obtained after labeling

For (3), there is a certain guarantee on the accuracy of labeling, but the cost is expensive

Images