APGEOG4440 M - Geoinformatics Remote Sensing II (Winter 2015-2016)
Lau Chun Wang
212577722
GEOG 4440
Lab: 001
TA: Harold
Prof: Dr. Tarmo Remmel
Observed Noise
In this section, basic information about the remote sensing image (A2Data.pix) will be provided, and the noise and artifacts within bands will be discussed as well. To begin with, there are 5 bands within the dataset which are derived from a Thematic Mapper (TM) Sensor for Landsat 4 and 5 satellites. Bands 1 to 5 represent blue, green, red, near infrared (NIR) and shortwave infrared (SWIR) respectively (Please refer to http://landsat.usgs.gov/about_landsat5.php). The uncorrected true color map as shown below is composed by inputting band 3 in red gun; band 2 in green gun; and, band 1 in blue gun. In order to present it in natural color, linear enhancement is applied.
From this map, different kinds of noise and artifacts are identified. They are usually caused by electronic errors when the satellite image acquisition is disturbed by the sensor or foreign artifacts, such as water vapor and floating sediments. At the first place, there is a layer of noisy speckles covering the whole map. The background inconsistency of the lake in the western area (i.e. dispersion of red dots over the lake) is a good proof of this statement. Moreover, one of the artifacts identified is a narrow black line running from east to west (or from west to east) at the northern part of the map (Artifact #1). Another artifact is a strip of sunlight observed at the southwest corner of the map (Artifact #2). These artifacts are highlighted in the following map. In order to be more specific, each band will be examined.
Band 1 (Blue): This map is composed by inputting band 1 into red, green, and blue guns at the same time with linear enhancement for better presentation. To begin with, high level of pepper and salt noise is obvious all over the image. Again, it is more obvious when we refer back to the noisy speckles dispersing above the lake in the western area. Rayleigh scattering, which is common in bands with shorter wavelengths, is one of the possible reasons attributing to the noise. Meanwhile, Artifact #2 can be observed at the southwest corner of the map.
Band 2 (Green): This map is composed by inputting band 2 into red, green, and blue guns at the same time with linear enhancement for better presentation. Similar to band 1, band 2 is also affected by the pepper and salt noise but at a lower level. Artifact #2 is also apparent the southwest corner of the map. Besides, Artifact #1 can be observed at the northern part of the map. Electronic array error, such as the failure of the sensor to detect electromagnetic radiation during image acquisition, can be a possible reason.
Band 3 (Red): This map is composed by inputting band 3 into red, green, and blue guns at the same time with linear enhancement for better presentation. Similar to both previous bands, band 3 is suffered from the problem of salt and pepper noise but at an even higher level. Radiation is possibly reflected by atmospheric particles or floating sediments. Such level of noise has caused serious distortion of data which is unfavorable for further data manipulation. Artifact #2 is also apparent at the southwest corner of the map.
Band 4 (NIR) and Band 5 (SWIR): The maps presenting band 4 is composed by inputting band 4 into red, green, and blue guns at the same time with linear enhancement for better presentation. Same method is applied to band 5 but with the input of band 5. The amount of distortion caused by the speckling or hazing is very limited in these 2 bands. Meanwhile, the presence of Artifact #2 in these 2 bands is extremely fainted as well. Therefore, correction will not be applied to them.
Please Click Image to Enlarge!