An Intelligent Cable Arrangement Detection Algorithm via Improved CNN Architecture and Edge Rectification Technology

This research proposes an intelligent cable arrangement detection algorithm integrating improved convolutional neural network (CNN) and edge rectification, focusing on the elongated object detection challenges in the automated cable deployment/retraction (D&R) control system. An edge rectification based on an optimized vision algorithm is proposed to simultaneously detect port-side cables and track winding-mechanism positions, transforming spatial coordinates into normalized percentage metrics for motion control. Two innovations introduce: 1) an enhanced CNN architecture with multi-scale features for linear-target detection precision, and 2) a boundary-correction algorithm eliminating positional errors while stabilizing control setpoints during layer transitions. Method validation via specialized dataset simulations confirms operational efficacy, achieving intelligent D&R through synchronized visual-adaptive control, enhancing autonomous platform operations.