船舶三维视觉图像模糊目标清晰处理仿真  

Simulation research on blur target clarity processing of ship 3D vision image

作　　者：李姗 LI Shan(Henan Fault-tolerant Server Engineering Technology Research Center,Zhengzhou 450046,China;Henan Polytechnic,Zhengzhou 450046,China)

机构地区：[1]河南省容错服务器工程技术研究中心,河南郑州450046 [2]河南职业技术学院,河南郑州450046

出　　处：《舰船科学技术》2023年第7期174-177,共4页Ship Science and Technology

摘　　要：为保证船舶目标的识别效果,需提高船舶图像目标的清晰度。因此,研究船舶三维视觉图像模糊目标清晰处理方法。该方法采用立体双目视觉在2个视点下观测同一个目标,依据2个观测面的模糊核路径和三维空间内的船舶运动路径相对应的原理,获取船舶模糊图像的三维信息;采用局部模糊探测方法,依据获取的三维信息分割模糊目标图像,并将分割结果输入残差聚集网络模型中,通过模型的处理获取清晰化高质量目标图像。测试结果显示,该方法清晰处理后的目标图像对比度、边缘强度以及图像方差3个指标的结果均在0.933以上,可有效完成图像模糊目标的清晰处理,获取清晰的目标结果。In order to ensure the recognition effect of the ship target,it is necessary to ensure the clarity of the ship image target.Therefore,the processing method of the ship 3D visual image fuzzy target clarity is studied.The method uses stereo binocular vision to observe the same object under two viewpoints,and obtains the three-dimensional information of the ship's blurred image according to the principle that the fuzzy core path of the two observation surfaces corresponds to the ship's motion path in three-dimensional space;The local fuzzy detection method is used to segment the fuzzy target image according to the acquired 3D information,and the segmentation result is input into the residual aggregation network model,and the clear high-quality target image is obtained through the processing of the model.The test results show that the contrast,edge intensity and image variance of the target image after clear processing by this method are all above 0.933,which can effectively complete the clear processing of the blurred image target,get clear target results.

关 键 词：三维视觉图像 模糊目标 清晰处理仿真 观测面 模糊核路径 三维信息 

分 类 号：TP392[自动化与计算机技术—计算机应用技术]