基于机器视觉特征强化的降质图像动态目标检测方法  

Dynamic object detection method for degraded images based on machine vision feature enhancement

作　　者：武狄 张国辉 张庆 Wu Di;Zhang Guohui;Zhang Qing(School of Computer&Information Engineering,Heilongjiang University of Science&Technology,Harbin 150022,China)

机构地区：[1]黑龙江科技大学计算机与信息工程学院,哈尔滨150022

出　　处：《黑龙江科技大学学报》2024年第4期648-654,共7页Journal of Heilongjiang University of Science And Technology

基　　金：黑龙江省省属高校基本科研业务费项目(7020000070226)。

摘　　要：针对降质图像中的动态目标特征不明显,易出现检测偏差的问题,提出基于机器视觉特征强化的降质图像动态目标检测方法。采用帧间阈值法确定图像中含有动态目标的区域,通过平滑、动态目标强化和背景抑制处理,引入SE-ResNeXt模块,基于挤压-激励操作后提取动态目标的特征信息,计算动态目标与背景间的速度和光流矢量,在速度平滑性、方向平滑性和颜色分量约束条件下,检测得到动态目标的准确位置。结果表明,所提的方法在准确率、召回率和F 1分数的最优参数组合下,PSNR和SSIM平均提高了3.41%和0.16%,提升了降质图像整体质量,实现了对动态目标的精准检测与识别。This paper proposes a dynamic target detection method based on machine vision feature enhancement to address the detection deviation caused by unclear dynamic target features in degraded images.The study consists of determining the regions in the image that contains dynamic targets through the inter frame theshold method to conduct smooth processing,dynamic target enhancement and background suppression processing;introducing the SE ResNeXt module to extract feature information of dynamic targets by squeezing and stimulating operations;calculating the velocity vector and optical flow vector between the dynamic target and the background,and detect the accurate position of the dynamic target under the constraints of velocity smoothness,directional smoothness,and color component.The results show that the proposed method selects the optimal combination of experimental parameters of accuracy,recall,and F 1 score.The overall quality of degraded images improves with the average improvements in PSNR by 3.41%and in SSIM by 0.16%,as which achieves the accurate detection and recognition of the dynamic targets.

关 键 词：机器视觉特征强化 动态目标检测 速度矢量 速度平滑性约束 颜色分量约束 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]