Infrared small target detection based on the tensor model

Small target detection is one of the crucial techniques of infrared search and tracking systems. A new infrared small target detection method based on the tensor model is proposed in this paper. Firstly, by cropping the real infrared images, small target samples are generated to a higher order tensor. Secondly, the N-mode SVD method is used to obtain the different characteristic matrices and the core tensor. Thirdly, the certainty image of existing a small target in each pixel location is measured by computing the distances between the eigenvectors of each characteristic matrix and the projection vectors of the each block in the test image. Finally, the small infrared target is detected by adaptively segmenting the certainty image. The results show that our method can enhance the small target effectively and can get a much better detection performance than many other traditional methods.