A Method of Infrared Image Non-uniformity Correction for Ground Scenes

Abstract

The invention belongs to the non-uniformity correction technology of an infrared focal plane detector, and proposes a method for correcting the non-uniformity of an infrared image in a ground scene. The steps of the non-uniformity correction method in this paper are as follows: use the correction coefficient corresponding to each pixel to correct and output the image; perform blind element judgment on the input image; calculate the motion parameters between the current frame image and the adjacent previous frame image ; Take the grayscale of the adjacent previous frame image as the expectation, update the correction coefficient of each pixel of the current frame image; take the grayscale of the current frame image as the expectation, correct the correction coefficient of each pixel of the adjacent previous frame image Coefficients are updated. This process is carried out synchronously with the input cycle of the image data to realize the iterative update of the correction coefficient.

Description

A Method of Infrared Image Non-uniformity Correction for Ground Scenes technical field The invention relates to a method for correcting the non-uniformity of an infrared image of a ground scene. Background technique During the production and use of infrared focal plane array detectors, many links will cause non-uniformity in the response between pixels, so it must be corrected for non-uniformity before it can be used. Commonly used correction methods include calibration correction methods and scene-based correction methods. The calibration correction method is to use one or more standard blackbody reference sources to calibrate the output value of each pixel of the detector to reduce non-uniformity. However, due to the time drift characteristics of the detector and changes in the use environment, the non-uniformity in the image will gradually increase, and even seriously affect the use of the product, so the calibration technology generally requires regular re-calibratio...

AI Technical Summary

Problems solved by technology

These methods have their own defects. For example, the time-domain high-pass filtering method and the neural network correction method require the scene to move randomly, otherwise ghost images will be left in the image or the image will be blurred; the Kalman filter method and the correction method based on motion estimation use Statistical information of frame blocks is used to obtain effective correction coefficients, the algorithm requires more memory, and the calculation complexity is larger, which is difficult to meet the requirements of real-time processing; the algebraic iteration method requires the design of a column or a row of reference pixels in the system, which is relatively difficult to implement. complex

At present, there is no scene-based correction method with good effect that can be used in practice, so a high-performance, adaptive dynamic inter-frame correction correction method is needed

Images