基于DSP的两帧差分和改进半因果弱小目标检测

摘要

针对工业现场镁熔液中弱小目标实时检测的难题,提出一种基于DSP的两帧差分和改进半因果支持域相结合的弱小目标检测新方法.该方法以DSP图像处理平台为核心,DSP集成开发软件Code Composer Studio 3.3为辅助,首先采用改进的半因果支持域模型对两帧原始图像进行背景预测,背景预测图像与原始图像相减得到残差图像;其次残差图采用两帧差分和自适应阈值分割处理得到二值图像;最后对二值图像形态学处理,获得真实弱小目标.实验结果表明:在相同实验条件下,该方法对序列图像的检测性能与已有文献进行对比,可以快速有效地检测到弱小目标,同时检测到的目标面积增大1.3倍,预处理时间减少34％,检测总时间减少20％,为工业现场镁熔液弱小目标实时检测奠定了基础.%Aiming at the problem of real-time detection for dim and small target in industrial scene,a new method of dim and small target detection which combined two frame difference and modified semi-causal support region based on DSP was proposed.Firstly,the two original images were estimated by using the modified semi-causal support domain model and the residual images were obtained by subtracting the prediction image from the original image.Secondly,the residual images were processed by two frame difference and adaptive threshold segmentation to obtain binary image.Finally,the real dim and small target was detected using morphological processing.Under the same experimental conditions,the detection performance of the proposed method was compared with that of the existing method.The experimental results show that this method can detect dim and small targets quickly and effectively,and meanwhile the detected target area is increased by 1.3 times,the pretreatment time is reduced by 34％ and the total detection time is reduced by 20％,which lays the foundation for real-time detection of dim and small target in the industrial field.