混合光照下复杂偏振光成像遮挡目标识别方法

Occlusion target recognition method for complex polarized light imaging under mixed illumination

遮挡现象会降低图像质量,淹没目标信息细节。为了更好地实现目标探测与跟踪,提出混合光照下复杂偏振光成像遮挡目标识别方法。在偏振成像系统中引入Stokes参量,构建混合光照强度矩阵计算偏振度与偏振角,采集目标图像;通过边缘检测与高斯平滑等过程,获取初始图像内目标边缘信息,去除图像噪声;根据普朗克辐射定理分析目标光谱特征,以特征向量形式描述选取的图像像素点深度值,经归一化处理后,去除冗余特征,确定被遮挡部分的关键特征点,引出关键线条;构建辅助坐标系,综合方位参数,将成像系统坐标系变换为图像平面坐标系,完成目标三维重构,有效识别遮挡目标。实验结果表明,所提方法的识别误差最高为0.125%,识别结果与实际目标的结果相似度高于0.97,识别耗时最高为7.22 s。

Occlusion will reduce the image quality and drown the details of the target information.In order to better realize target detection and tracking,an occlusion target recognition method based on complex polarized light imaging under mixed illumination is proposed.The Stokes parameter is introduced into the polarization imaging system,the mixed illumination intensity matrix is constructed,the degree of polarization and polarization angle are calculated,and the target image is collected;Through the processes of edge detection and Gaussian smoothing,the target edge information in the initial image is obtained and the image noise is removed;The spectral features of the target are analyzed according to Planck radiation theorem,and the depth values of the selected image pixels are described in the form of feature vectors.After normalization,the redundant features are removed,the key feature points of the occluded part are determined,and the key lines are derived;Construct the auxiliary coordinate system,integrate the azimuth parameters,transform the imaging system coordinate system into the image plane coordinate system,complete the three-dimensional reconstruction of the target,and effectively identify the occluded target.The experimental results show that the maximum recognition error of the proposed method is 0.125%,the similarity between the recognition result and the actual target is higher than 0.97,and the recognition time is up to 7.22 s.