Data Integration through IHS Color, PCA and Arithmetic Operation

Introduction

        The objective of data integration is to obtain more information in a display than is possible from any one data source, thereby increase the analytic value of each source. SAR is often analyzed process in this way, taking advantage of its unique capability in the analysis of topography, surface roughness, and moisture content. In this assignment, SAR data will be integrated with optical satellite data (Landsat TM), thematic data, and continuous imagery. A variety of data integration techniques will be performed, including principle Component Analysis, Intensity Hue Saturation, and Arithmetic. The utility of these methods by themselves and in combination will be explored.

Data Integration- Requirements

1. Evaluate the SAR image for speckle. Based on the theory discussed in class in class, run appropriate filter(s) on this image to reduce speckle while maintaining sharp edges. Utilize your best filtering results for the remaining analysis.
2. (I-H-S) Perform an RGB I-H-S analysis on the optical dataset the purpose of image enhancement. Base on the theory discussed in class; compare the results to an enhanced RGB composite (Hint: You have done this correctly if it looks very similar).
3. Integrate the SAR and optical dataset using an I-H-S transformation with SAR substituted as intensity. Compare this with and arithmetic combination of SAR and optical data. Note that this can also be accomplished using the RGBI transformation (ERMapper can automatically do a simple IHS transformation –it get saturation from the RGB).
4. Integrate the SAR with a thematic dataset using an I-H-S transformation. Experiment with using a synthetic channel for saturation. Determine the effects of varying the saturation level (eg. 50, 150, 200). Compare this with use of an optical channel(s) for saturation.
5. Integrate the SAR with a continuous dataset using an I-H-S transformation.

PCA

6. Perform a principle component analysis (PCA) on the optical dataset. Compare the results with a PCA with the SAR included as one of the channels. Manipulate the statistic to determine the loading in each case (include in the appendix of your report).
7. Use the PCA results from the optical dataset create RGB composite enhancement. Integrate three of the components with the SAR using I-H-S.
8. Use the PCA that includes SAR (number 6), combine with either your thematic or continuous dataset to obtain the maximum possible integration in I-H-S. Examine the resulting display to interpret its meaning.

Available Data

        In this assignment, I have select the Natuna Island, which I would like to present my assignment by using the data from RADARSAT, which have details following: 1) Geographical Area: Natuna 2) orbit: 4272 Ascending, 3) Beam Mode: SAR Standard 6 Beam, 4) Product Type: Path Image (SGF), 5) Pixel Spacing: 12.500 m. and Optical Data, which have details following: 1) Geographical Area: Natuna, 2) Sensor: Landsat 5 TM 3) Resolution 30 m, Date of Acquisition 05 July 1997 for the location that I have interested will be shown in the procedure that we have to cut or make sub-image and then examine carry on.

Background

What is data Integration, combining different data sets to produce images that contain more detailed information than can be displayed from the original image alone/Data integration in these procedures we are interested in the process how to perform by using to combine image data for a given geographic area with other geographically referenced data sets for the same area. These other data sets might simply consist of image data generation on other dates by the same sensor or by other remote sensing system. Frequently, the intent of data merging is to combine remotely sensed data with other source of information in the context of a GIS. For example, image data are often combined with soil, topographic, ownership, zoning, and assessment information.

In the Hue Saturation Intensity (HSI)

     In the Hue Saturation Intensity (HSI) Color system, different colors are characterized by three measurable characteristics of a color:

Hue–The main attribute of a color that distinguishes it from other colors in the spectrum. Hues are what you see in a rainbow, and are what we commonly think of as "color" (red, yellow, green, and so on).

Saturation–The amount of grey in a color or color "purity." Colors with high saturation are said to be pure or vivid. Colors with low saturation (much grey) are pastel or dull colors. Completely desaturated colors are grey, no matter what the hue.

Intensity–The relative brightness of a color. Colors with high intensity are bright, and colors with low intensity are dark.

      The HSI color system is characterized as a "perceptual" color system because it provides a more intuitive means of manipulating color than the RGB (electronic) color system. ER Mapper implements the HSI color system by means of the Color Mode named Hue Saturation Intensity, and separate algorithm layer types for Hue, Saturation, and Intensity.

Why is IHS color space useful for image display?

        IHS allows an image to be displayed with more feature detail, which the RGB color space could allows only 3-band channels to displayed, while 9-band channel may be used in an IHS image, this is the reason that why we have to perform them.

Arithmetic Modeling

         Arithmetic modeling allows for varying proportions of different types of images to be displayed and an the equation is specified to separate the image display into varying proportions of the image types and also useful because the unique qualities of one image type may be displayed with the unique qualities of one or more other images.

Principle Component Analysis (PCA)

         Principal Component Analysis (PCA) basically is a statistical analysis. The image data sets that perform data integration though PCA, which PCA use for data enhancement, data reduction and integration. Very useful for highly correlated data for reducing data redundancy and correlation between bands that the output and components are linear combinations of the original data the reason that why we have to perform a Principal Component Analysis due to we would like the new components that could used for image display rather than the original layers and also for reducing inter-band correlation, highlights low contrast features that the PCA have had good for Change detection analysis that we will be studied in the assignment number2.

Previous                 Next