A Pixel Missing Patch Inpainting Method for Remote Sensing Image

Image inpainting is an important research field in remote sensing image processing. Pixel missing patches on remote sensing image due to clouds and stripes were often processed through traditional synthesis and image inpainting techniques or multi-spectral and multi-temporal based methods. These methods have some limitations in practical applications. In this study, a new method was developed to restore pixel-missing patches on remote sensing images. It uses a corresponding recent image (reference image) as a Look-up table and fills up pixel missing patches using the non-missing pixels in the pixel missing image. The restoration effect of this new method was tested using two experiments. The restored image was almost identical to the original one visually. This new method is applicable when the reference image and pixel missing image are acquired from same or different sensors. It is also insensitive to the precision of atmospheric corrections for the reference and pixel missing images.