Friday, September 14, 2007

A Brief Knowledge on Remote Sensing Analysis

Its been more than 8 months since I graduate from my Master degree in The University of Melbourne. Today I open my computer and found one file of my assignment on Remote Sensing. I think it would be useful to spread my knowledge in this subject so I wrote down basic methods that are commonly used in Remote Sensing analysis.

Enjoy !!

Single Band
Band 2 (Green) is useful for vegetation discrimination, water quality and chlorophyll concentration. Band 3 (red) is used to analyze the amount of chlorophyll in the vegetation. Since leaf chlorophyll absorbs red band energy, the lower the reflected energy, the greater chlorophyll amount.
Near-infrared band (band 4) picks up the portion of the reflected light spectrum created by the plant’s mesophyll leaf tissue. Consequently vigorously growing healthy vegetation has low red-light reflectance (due to its chlorophyll) and high near-infrared reflectance (due to its total biomass). Band 5 (shortwave infrared) is useful to analyze moisture levels in soil and monitoring plant vigor.

Ratio Band
Ratio band is an analysis to combine two band colors to produce a ratio image depends on the analysis mean. For example, for vegetation analysis is common to use the ratio of band 4/band3.

VI (Vegetation Index) and NDVI (Normalised Difference Vegetation Index)

The NDVI value is influenced by the chlorophyll content of the vegetation and due to this, the NDVI image looks similar to the vegetation ratio.
Most suitable for vegetation applications are the band combination NIR, R, G = RGB and the Vegetation Ratio (NIR/R) and Normalised Difference Vegetation Index (NDVI = (NIR-R) / (NIR+R))

PCA (Principle Components Analysis)
This technique could identify and remove redundancy in multispectral image data (between spectral bands and between images that are acquired on different dates).
PCA answers the challenges to monitoring change over with multi date images to identify the relatively small proportion of pixels that actually changes. PCA’s first component image will show features that did not change over time while the other components will highlight features that exhibit different degrees of change.

Image Enhancement
Image enhancement refers to a number of image processing procedures that improve the visual interpretability of an image, done by applying algorithms that changes the contrast, brightness, sharpness and color rendition of features. Enhancement does not contain more information that the original data.

Filter is used to verify nearby difference colors. Filters could enhance the edges and other frequency information while it eliminates gradual trends in the image. Using the “high pass” option is more useful because it produce more contrast image.

Image Classification
Image supervised classification defines an area in the image that are represents of each information class and has the computer generate to the image. On the other hand, unsupervised classification is done manually by an interpreter based on pixel values, sensor viewing angle, and health and growth stage.

No comments: