Research on polarization imaging detection method for moving object in complex scenes

Aiming at solving detection problem of moving objects in complex scenes that have the features of low contrast and low Signal-to-noise ratio, a polarization imaging detection method based on stable principal component pursuit was proposed. Firstly, a matrix was composed by pre-processed polarization image sequences. Based on correlation of image sequences, stable principal component pursuit mathematical model was build. The matrix could be decomposed to a low-rank matrix, a noise matrix, and a sparse matrix containing object information. Secondly, sparse matrix was achieved using augmented Lagrange multiplier method when the value of object function was sum of nuclear norm of the low-rank matrix and 1 norm of the sparse matrix was the smallest. Finally, Markov random field was used to remove noise in sparse matrix. The experimental results show that the adaptability to complex scenes and the accuracy rate of this method is better than other detection methods.