Multi object detection method in ship navigation images based on visual communication
In response to the complexity of the marine environment and the diversity of ship targets,a multi object de-tection method based on visual communication in ship navigation images is studied to timely discover potential safety haz-ards.Convert ship navigation images from RGB color space to CIE Lab mode space,and apply an improved HFT(Hyper-complex Fourier Transform)model to effectively extract multi-target salient regions of ships.Using weighted processing and Otsu algorithm to partition multi-layer salient regions,and determining multi-target candidate regions for ships based on pri-or information.Using fuzzy C-means clustering algorithm to segment candidate regions and achieve accurate detection of multiple targets on ships.The experimental results show that this method can accurately detect multiple targets of ships in various complex environments such as cloud and fog coverage,ocean clutter,and ship wake,and has high robustness and practicality.
visual communicationship imagesmulti object detectionsignificant areasIs clusteringcandid-ate region
NETL
NSTL
万方数据
