The invention belongs to the technical field of
computer vision and electronic
engineering, and discloses an
instrument control drawing information intelligent extraction method and
system, and the method comprises the steps: detecting components in a drawing, outputting a component bounding
box set Yv0, carrying out the instance segmentation of a text region in the drawing based on a
Mask R-CNN model, generating a text
mask set Rv0, and recognizing the text content. The method comprises the following steps: generating a text semantic set, extracting a connecting line skeleton through
edge detection, morphological closed operation and direction filtering, executing iterative
mask constraint optimization, obtaining an optimized detection result, establishing an association relationship between a component and a text, traversing intersection points and coverage points of the connecting line skeleton in a horizontal / vertical direction, and generating a component topological connection diagram; the
system comprises a visual acquisition module, a multi-task detection module, a dynamic optimization module, an association reasoning module and an output module. According to the method, the accuracy of element recognition and correlation reasoning in the complex drawing can be improved, and meanwhile, the robustness and generalization ability of the
system are kept.