Sparse Representation for Infrared Dim Target Detection via a Discriminative Over-Complete Dictionary Learned Online

Abstract

It is difficult for structural over-complete dictionaries such as the Gabor function and discriminative over-complete dictionary, which are learned offline and classified manually, to represent natural images with the goal of ideal sparseness and to enhance the difference between background clutter and target signals. This paper proposes an infrared dim target detection approach based on sparse representation on a discriminative over-complete dictionary. An adaptive morphological over-complete dictionary is trained and constructed online according to the content of infrared image by K-singular value decomposition (K-SVD) algorithm. Then the adaptive morphological over-complete dictionary is divided automatically into a target over-complete dictionary describing target signals, and a background over-complete dictionary embedding background by the criteria that the atoms in the target over-complete dictionary could be decomposed more sparsely based on a Gaussian over-complete dictionary than the one in the background over-complete dictionary. This discriminative over-complete dictionary can not only capture significant features of background clutter and dim targets better than a structural over-complete dictionary, but also strengthens the sparse feature difference between background and target more efficiently than a discriminative over-complete dictionary learned offline and classified manually. The target and background clutter can be sparsely decomposed over their corresponding over-complete dictionaries, yet couldn't be sparsely decomposed based on their opposite over-complete dictionary, so their residuals after reconstruction by the prescribed number of target and background atoms differ very visibly. Some experiments are included and the results show that this proposed approach could not only improve the sparsity more efficiently, but also enhance the performance of small target detection more effectively.

Infrared small moving target detection algorithm based on joint spatio-temporal sparse recovery

“…In this paper, the structural Gaussian spatio-temporal over-complete dictionary is adopted to distinguish target spatio-temporal atom from background spatio-temporal atom automatically [18], and a target spatio-temporal atom and a background spatio-temporal atom are shown in Fig. 3.…”

Small target detection using edge-preserving background estimation based on maximum patch similarity

“…[17][18][19] A traditional LCM for small infrared target detection was proposed 3 based on HSV. However, the region with high brightness PSEN which usually exists in the infrared image also have contrast differences with its surrounding areas; therefore, the LCM may leads to a high false alarm and a low signal-to-clutter ratio (SCR) 20 of the image. In order to deal with this problem, we propose an ILACM in which the structural feature of the small target is adequately considered.…”

Infrared LSS-Target Detection Via Adaptive TCAIE-LGM Smoothing and Pixel-Based Background Subtraction

“…Gao et al [7] built a co-detection model based on nonlinear weight and entry-wise weighted robust principal component analysis (RPCA), which can extract real targets accurately and suppress background clutters efficiently. Li et al [8] proposed an infrared dim target detection approach based on sparse representation on a discriminative over-complete dictionary, which can not only capture significant features of background clutters and dim targets, but also strengthen the sparse feature difference between the background and target. Kim et al [9] analyzed the characteristics of regional cluster and removed the false detection by means of spatial attribute-based classifications, the heterogeneous background removal filter, and temporal consistency filter.…”