Abstract: The high precision nighttime lights data can effectively reflect the characteristics of human space activities.The traditional SNPP-VⅡRS nighttime lights images denoising method is easy to ignore the newly added or disappeared effective light,and has not dealt with the light outliers.With the passage of time,the denoising error will gradually increase.Therefore,this paper proposes to filter the outliers and background noise of SNPP-VⅡRS nighttime lights images using a combination of median filtering and low threshold denoising,and the correlation between the total amount of nighttime light after denoising and GDP is analyzed and the accuracy is verified.Experimental results show that the correlation between the total amount of nighttime light collected by median filtering and low threshold denoising and GDP is superior to the traditional method,and the precision of the brightness estimated by the model established by this method is higher,reflecting that the method has a higher denoising precision.

Key words: SNPP-VⅡRS, nighttime lights, median filtering, low threshold denoising