Feature extraction for rescue target detection based on multi-spectral image analysis

In order to solve the problems with rescue target detection under complex sea environment, the research on object detection on the sea surface based on multi-spectral analysis and machine vision is presented in this paper. Firstly the difficulties and features of weak target detection and its complex sea background are discussed and the multi-spectral image acquisition and pre-processing method suitable for this task are proposed. Then the spectral features of targets and sea surface are extracted by analyzing visual light spectra and near-infrared spectra, which can distinguish the ships or survivors from islands or sea wave. The main purpose of this task is to detect lifeboats or survivors on the sea by analyzing the multi-spectral information, which can propose a target detection method based on image analysis and machine vision to find the survivor in the case that a marine casualty occurs, and then help the rescuer find the survivor more efficiently.