Bad Pixels Detection And Compensation For Longwave Cooled Infrared Focal Plane Imaging System

Infrared imaging system with high resolution, good for hiding and not susceptible toelectronic interference, etc. are more and more used in industrial, civil, military and otherfields. The core components of the infrared imaging system is infrared detectors, In1990s,China are able to self-developed infrared imaging system, but the core components of theinfrared detector is rely on imports. As in recent years, vigorously developing countries onthe basis of physical, material and other disciplines, to promote the development ofdomestic high-speed infrared detectors, China is now able to self-developed andproduction cooled shortwave infrared focal plane array detectors, its performance hasreached international advanced level. However, independently developed by China than inother countries there is a considerable gap between the performances applied to spaceexploration in the long-wave infrared focal plane cooled. The purpose of the article is toimprove the image quality of the domestic long-wave cooled infrared focal plane detectorby digital image processing methods.In this paper, the use of three-dimensional noise model of domestic long-wave cooledinfrared focal plane imaging system noise analysis, found that the biggest factor affectingimage quality ranks noise, mainly on the non-uniformity of the FPA. When the detectorpixel non-uniformity reach extreme cases, it will become the response of nonlinearsingular points, these points more than the performance of the infrared image with thescene changes fixed bright, dark spots and flicker over the scene change point These dotsare called blind pixel. The blind pixel serious impact the quality of infrared imagingsystem imaging, infrared imaging system will also affect non-uniformity correction,therefore before the image non-uniformity correction, the blind pixels must be removed.By analysis of generation mechanism and the characteristics of the blind pixels, wedivide blind pixels into four types: detector and readout circuit is not connected, whichhaven’t photoelectric response and it export high voltage; large connectivity resistorbetween detector and readout circuit, which have photoelectric response and export highvoltage; the blind pixels which generate by the PN junction damaged, export low voltageand haven’t photoelectric response; The photodiode impedance drop formation Blindpixels, which have photoelectric response and export low voltage.According to the characteristics of the blind pixels, this paper proposes an algorithmwhich combining of the fixed blind pixels correction and dynamic blind pixel correction inthe infrared imaging system. For the blind pixels generate by PN junction damage andNon-well-connected between detector and the readout circuit, due to its photoelectricresponse is a fixed value, using the fixed blind pixel correction algorithm; For the other kind of blind pixels, because of their response will change follow the scene changes, sousing the dynamic blind correction algorithm. Through the analysis of experimental data,the proposed algorithm not only has better treatment effect, and it’s processing time isvery short, very convenient hardware logic, to meet the requirements of today’s infraredimaging technology for high resolution, high frame-rate real-time infrared imageprocessing, and thus has good practical value.