Automated design parameter extraction and restoration from 2D propeller drawings

Abstract

The shipbuilding industry increasingly needs 3D propeller models from 2D drawings for repair, retrofitting, and energy-saving device (ESD) analysis. However, clients often provide only 2D drawings due to security, making manual information extraction for propeller models time-consuming, labor-intensive, and prone to errors. This highlights the need for automated, accurate extraction techniques. This study proposes a line detection and information extraction method to obtain design parameters from 2D propeller drawings. The method converts PDF drawings to images, preprocesses them, and then uses a path-finding algorithm to detect lines and extract information. This extracted data is converted into design parameters like rake, skew, chord length, camber, and thickness through offset data acquisition. Applying this method to propeller drawings significantly reduces time and effort compared to manual work, greatly improving efficiency and restoration accuracy. The method effectively detects complex and overlapping lines, and the quantitative accuracy of the extracted design parameters has been validated, with most parameters showing less than 1 % error.