SARTEX

Description

SARTEX creates a set of texture images from a single channel in an input image. The radar-specific measures are based on pixel-value statistics in a window surrounding each pixel. If necessary, you can use the extracted texture measures as input features with CATALYST Professional classification algorithms.

Parameters

Name	Type	Length	Value range
Input: Detected-SAR channel *	Raster port	1 - 1
Normally distributed intensity	Boolean	0 - 1	TRUE | FALSE
Normally distributed amplitude	Boolean	0 - 1	TRUE | FALSE
Lognormal distributed intensity	Boolean	0 - 1	TRUE | FALSE
K-distributed intensity	Boolean	0 - 1	TRUE | FALSE
Output: Texture measure channels *	Raster port	1 - 4
WindowXSize	Integer	0 - 1	3 - 101 Default: 7
WindowYSize	Integer	0 - 2	3 - 101 Default: 7
Area Mask: Area mask	Bitmap port	0 - 4	Xoffset, Yoffset, Xsize, Ysize
Port Settings: Resample Mode: Method of resampling	Raster port	1 - 1024	Nearest | Bilinear | Cubic Default: Nearest
Image Format	String	0 - 1	Amplitude | Power | Decibel Default: Amplitude

* Required parameter

Parameter descriptions

Input: Detected-SAR channel

The input SAR channel to analyze for texture measure.

The channel must be detected data, and not complex numbers.

Normally distributed intensity

To calculate texture with intensity distributed normally, select this check box.

This is a ratio of the mean of squared intensity to the mean intensity squared. It is applicable when pixel intensity has a Gaussian distribution.

For more information about this measure, see Algorithm.

Normally distributed amplitude

To calculate texture with amplitude distributed normally, select this check box.

This is a ratio of mean intensity to the squared mean amplitude. It is applicable when pixel amplitude has a Gaussian distribution.

For more information about this measure, see Algorithm.

Lognormal distributed intensity

To calculate texture with intensity distributed log-normally, select this check box.

This is a difference of the mean value of the squared intensity logarithm and the square of the mean intensity logarithm. It is applicable when pixel intensity has a log-normal distribution.

For more information about this measure, see Algorithm.

K-distributed intensity

To calculate texture with intensity K-distributed, select this check box.

This is a normalized log measure of texture. It approximates K-distribution when there are a large number of looks in the SAR image.

For more information about this measure, see Algorithm.

Output: Texture measure channels

The list of output channels to contain the calculated texture measures.

The input channel cannot be used as an output channel. The output channels must be 32-bit real.

WindowXSize

The horizontal size, in pixels, of the window to use to extract the texture measures for each input pixel.

The value you specify must be an odd integer ranging from 3 through 101. The default value is 7.

WindowYSize

The vertical size, in pixels, of the window to use to extract the texture measures for each input pixel.

The value you specify must be an odd integer ranging from 3 through 101. The default value is 7.

Area Mask: Area mask

The bitmap that defines the area of the input channel to process.

Port Settings: Resample Mode: Method of resampling

The resampling method to apply.

Image Format

The image units of the input SAR image.

Amplitude is the square root of power. Most radar images are stored in amplitude format.

The input values will be converted to the format required by the selected texture measures (power and amplitude).

This parameter is optional.

Details

SARTEX creates a set of texture images from a single channel in an input image. The radar-specific measures are based on pixel-value statistics in a window surrounding each pixel. If necessary, you can use the extracted texture measures as input features with CATALYST Professional classification algorithms.

Texture is an important characteristic used to identify objects or regions of interest in an image. Unlike spectral features, which describe the average tonal variation in the various bands of an image, textural features contain information about the spatial distribution of tonal variations within a band.

The SAR-specific texture measures that SARTEX applies account for radar-image formation and statistical properties of radar speckle; therefore, it is essential that filtering has not been applied previously on the input channel.

The output images contain the raw texture measures, which may vary in ranges of values. To avoid loss of important information, always save in 32-bit real channels. After assessing the distribution of texture-measure values by running the HIS algorithm, you can scale the images to the range you want by running the SCALE algorithm.

With the Area mask parameter, you specify the area within the input channel to process. Only pixels under the mask area receive output textures. The remaining output pixels are not modified.

SARTEX cannot process pixels on or near the edges of the image, because all the required pixels are not within it. Therefore, when an entire image is processed with a rectangular window that is x-pixels by y-lines, the first and last (X+1)/2 pixels per line will be the same and the first and last (Y+1)/2 lines in the output image will be the same. Each will contain a replicated value of the closest processed pixel. Avoid these regions when selecting training data for supervised classification. If a partial image is processed, the input image buffers are extended towards the margins as far as possible, so that all pixels in the selected area receive computed textures; however, pixels closer than a half-window to the image edge will contain replicated values.

Texture measures extracted by SARTEX complement measures extracted by running the TEX algorithm, based on a gray-level co-occurrence matrix (GLCM), and with the HISTEX algorithm, based on the histogram of pixel values in a window. To conduct additional analysis, such as standard segmentation and classification of the image, you can combine the three types of measures. In this case, the best and most effective combination depends on land-cover characteristics, and may require some testing to establish. When classification methods account for properties of radar speckle, use only the texture measures extracted with SARTEX.

The optimum window size depends on image characteristics (terrain roughness, land-cover type, imaging parameters). With a large window, blocky artifacts may appear in texture images if a small, bright target is enclosed fully by the window. In most cases, use a window size that is small to moderate.

Algorithm

With SAR images, a single pixel conveys little information about the underlying ground cover. Interpretation occurs only by looking at the values of many pixels. The statistics that apply best to a given area depend on the actual image-formation process.

Texture measures computed by SARTEX are based on Chapter 8 of the book by Oliver and Quegan cited in References.

The measures are extracted in a window with pixels (K) and lines (L), sliding over successive pixels of the image.

Depending on the texture measures and the format of the input image you specify, the input pixel values may need to be converted to radar power (intensity), and the VA measure converted to amplitude. The conversion is performed according to the following expressions.

Amplitude (AMP) is converted to power (POW) as follows:

POW = AMP^2

Power in decibels (DB) is converted to power POW as follows:

POW = EXP10( DB / 10.0 )

Power (POW) is converted to amplitude (AMP) as follows:

AMP = SQRT( POW )

If the pixel power is zero or negative, the pixel is excluded from the accumulated sums for the current window. If all pixels in a window are excluded, the pixel is assigned a zero value for all selected texture measures.

References

Oliver, C. and S. Quegan (1998). Understanding Synthetic Aperture Radar Images. Artech House, Boston, London.