论文部分内容阅读
MODIS 1B影像中存在横向和纵向条带,基于均衡化曲线补偿技术,提出了一种快速检测和去除条带的算法,给出了条带检测判定参数,即:如果A帧和B帧的10组传感器校正系数C_i值分别位于0的两侧,且ABS(C_i)>0.008,则存在宽条带;如果10组传感器的校正系数MAX(C_i)>0.05时,则存在单行条带。去除宽条带时,需以当前列为首列,从其后连续的100列影像中提取各传感器相对当前列的校正系数。结果表明,条带检测模型能有效识别单行条带和宽条带,多行条带需目视识别,单行条带和宽条带去除效果明显好于NASA网站公布数据,多行条带虽未彻底去除,但处理后的影像质量得到明显改善。
There are horizontal and vertical bands in the MODIS 1B image. Based on the equalization curve compensation technique, a fast detection and removal algorithm is proposed. The parameters of the detection of the bands are given, that is, if the A and B frames of 10 The group sensor calibration coefficient C_i values are located on both sides of 0 respectively, and the ABS (C_i)> 0.008, there is a wide band. If the correction coefficient MAX (C_i) of the 10 groups of sensors is larger than 0.05, there is a single row of stripes. When removing the wide strip, the current column is taken as the first column, and the correction coefficient of each sensor relative to the current column is extracted from the subsequent 100 consecutive images. The results show that the strip detection model can effectively identify single-line strips and wide strips, and the multi-line strips need to be visually recognized. The single strip removal and wide strip removal are obviously better than those published by the NASA website. Thoroughly removed, but the image quality after processing has been significantly improved.