Multi-FAN: multi-spectral mosaic super-resolution via multi-scale feature aggregation network

This paper introduces a novel method to super-resolve multi-spectral images captured by modern real-time single-shot mosaic image sensors, also known as cameras. Our contribution is twofold. Firstly, we from rather than cubes, which helps take into account the spatial offset of each wavelength. Secondly, introduce an external multi-scale feature aggregation network (Multi-FAN) concatenates maps with different levels semantic information throughout super-resolution (SR) network. A cascade convolutional layers then implicitly selects most valuable generate image. merged generated SR produce quantitatively superior We apply our Multi-FAN RCAN (residual channel attention network), state-of-the-art algorithm. show that improves both quantitative results (PSNR and SSIM) inference time.