Destriping method for mid-infrared image
中波红外图像条纹处理方法

An image obtained by a satellite in orbit may reflect a variety of inaccurate observations and different forms of interference, such as image rotation, the bow effect, and non-uniform stripes phenomena. These are attributed to the inherent characteristics of the Earth. These types of redundancy and stripes can be foreknown, and the image can be restored using design optimization, the satellite orbit ephemeris and image processing methods. The restoration of a periodic or aperiodic interference newly introduced in orbit requires the integrated use of multiple data sources, prior knowledge on the principle of interference, and appropriate image processing methods. Meanwhile, the authenticity of the original radiation information should be preserved for follow-up quantitative studies. In the present study, a new destriping method was introduced for processing of Mid-Infrared images. Using on-board calibration data and analysis of the stripes phenomenon and mechanism, a new spatial domain compensation algorithm was explored for the removal of image interference stripes. Results show that the image quality can be substantially improved by the proposed method. When compared with the frequency domain filtering algorithm, it is better in terms of both the dynamic range index and radiation information preservation.