An algorithm of removing highlight target's smear in space observation image

Abstract

Smear effect is a “fact of life” when using frame transfer type charge-coupled devices (CCDs) for image or video sequence acquisition. Usually, CCD smear effect can be decreased greatly by employing certain measures. But smear effect can seriously disturb the dim targets detection in Star Observation Image (SOI). In order to realize automatic removing highlight target's smear in SOI, an automatic de-smear system is established. Algorithms such as Gaussian noise distribution parameters estimation, smear detection and gray level correction etc. are investigated. First, Gaussian noise's distribution parameter of SOI is estimated with histogram least square curve fitting. Subsequently, utilizing smear features in observation image, a smear detection algorithm based on statistical information is proposed. Then, after the smear position is determined, contaminated pixel's gray level is corrected. Finally, a set of de-smear system in star observation image has been developed with Visual Studio 2005. Experimental results indicate that with the SOI of 16 Bits and 1024 pixels×1024 pixels, single frame image processing time is about 300 ms. Smear effects are well corrected. And useful information of stars and target has not been destroyed. The processed SOIs can satisfy the demands of stability, reliability and precision for dim target detection.