Infrared object detection via patch‐tensor model and image denoising based on weighted truncated Schatten‐ <i>p</i> norm minimization

The nuclear norm minimization (NNM) is a special non-convex rank convex relaxation scheme for image denoising and object detection that requires background subtraction. Considering excessive shrinkage of components equal treatment different components, NNM extended to the weighted Schatten-p (WSNM) with weights assigned singular values. In this paper, multi-channel truncated WSNM model based on optimization framework proposed RGB colour denoising. On basis noise intensities non-local self-similar patches each channel itself, improved significantly by methods superposition truncation. Meanwhile, it can be generalized tensor space employed infrared imaging target spatial-temporal first time. And efficient alternating direction multiplier-based algorithms are developed solve accuracy algorithm effectively choosing an adaptive threshold. Extensive experiments real data verified method state-of-the-arts effectiveness.