Segmentation Based Sensor Fusion of Distorted Multiframe-Images

This paper presents a fast method to integrate image information from multiple frames to produce a single image-frame with super deep resolution. By measurement of the relative local sharpness with a local gradient operator, an in focus and out of focus region feature is computed. This feature is incorporated into thresholding segmentation schemes to reconstruct a single sharp image. Edge bleeding is discussed as a main source of error introduced by the sharpness measure and afterwards tackled by the segmentation algorithm. The proposed method is presented showing some image examples. The multiframe solution shows preferable results both from the point of view of image quality and computational complexity to adaptive identification restoration algorithms.