Research On Methods For Infrared Maritime Target Detection

The fast search for maritime targets is crucial for our country to cope with emergent maritime issues.Because the infrared imaging technology has excellent performance on capturing figures in complex conditions,the infrared maritime target detection has been a research hotpot in the area of fast search for maritime targets for recent years.In this situation,based on a thorough investigation on both overseas and domestic related works,and a deep research on infrared maritime target detection algorithms,an infrared maritime target detection algorithm structure working on complex sea conditions was proposed in this dissertation.This structure comprised algorithms that could work on normal sea conditions,strong path-radiation conditions and strong backlighting conditions.First of all,for normal sea conditions(like mild or strong ocean waves),a detection mechanism of 'Preprocessing + Single-Frame Detection + Multiframe Judgment' was proposed.An image preprocessing method based on local peak singularity analysis was used to suppress image background and enhance targets at first.Then,a single-frame detection(SFD)module was used to extract suspected targets from images.This single-frame detection module was composed of SFD algorithm based on wavelet transform,SFD algorithm based on visual attention model,and SFD algorithm based on histogram transformation.After this,all suspected targets would be judged by an antivibration pipeline-filtering algorithm.This judgment algorithm used interframe block matching to resist interframe jitter and exploited the continuity of target's interframe trajectory to extract real targets.Secondly,for strong path-radiation conditions(like heavy sea fog),a target detection method on the basis of gray region division was designed.Its basic idea was to divide different image gray regions via global threshold at first,and then,extract suspected targets from different gray regions through SFD module.After combining these suspected targets,the multiframe judgment method would be used to detect all real targets.Finally,for strong backlighting conditions,a target detection method based on spatiotemporal saliency map was proposed in this dissertation.This method deeply analyzed target's spatial and temporal saliency to effectively distinguish targets from background clutters.To verify the target detection performance of the infrared maritime target detection algorithm structure proposed here,the author of this dissertation captured real maritime image sequences both onshore and aerially.Based on 28 captured image sequences(36424 single images in total),experiment results showed that this algorithm structure had superior capability of detecting infrared maritime targets in complex sea conditions.