论文部分内容阅读
基于场景的盲元检测与补偿算法能有效地克服固定和随机盲元造成的图像亮点与暗点,提高探测器成像质量。通过将待检测图像分解为盲元目标和一般背景,建立迭代修正的背景预测模型提高预测准确度,并根据盲元和正常像元的残差能量统计差异完成单帧图像盲元检测。依据盲元缓变特征,利用建立的背景预测模型和盲元检测结果设计补偿算法,实现了单帧和序列图像的快速盲元补偿。结果表明:该算法对孤立和连续盲元均能准确检测,校正效果好,是一种有效、实用和易于软、硬件实现的在线检测及补偿算法。
The blind detection and compensation algorithm based on scene can effectively overcome the bright and dark spots caused by fixed and random blind elements and improve the imaging quality of the detector. By decomposing the image to be detected into blind target and general background, an iteratively revised background prediction model is established to improve the prediction accuracy, and the blind detection of single frame images is completed according to the statistical difference of residual energy of blind pixels and normal pixels. According to blind trait slowly changing feature, using the established background prediction model and Blind element detection results to design compensation algorithm, the fast blind element compensation of single frame and sequence image is realized. The results show that the algorithm can detect both isolated and continuous blinds accurately and has good correction effect. It is an effective, practical and easy software and hardware detection and compensation algorithm.