• Layer functionalities
• Filters
• Description
• Enhancements
• Spatial filters
• Colour adjustment
• Edge detection
• Masks
• Histogram
• Description
• Preferences
• Radiometric enhancement
• Save View as image
• Clipping layers
• Zoom to raster resolution
• Automatic vectorization
• Analysis view

Layer functionalities

Filters

Description

Filtering is a process by which we can enhance images. gvSIG can filter images through a variety of filtering methods. In the upper left part of the Filter dialog, the filters are grouped by type (1). By double-clicking one of the filters or by clicking on the "Add Filter" button on the bottom left, the filter will be added to the list of filters in the lower left part of the Filter dialog. All filters in the filter list will be applied in the preview. If you want to remove a filter from the list, you can either double-click on the filter or click on the "Delete filter" button. The filters in the list will be applied to the image in the order that they appear. Keep in mind that the order in which the filters are applied will affect the result, and changing the order of the filters may change the output.

In the middle of the dialog window are the controls of the selected filter (2). When changing the controls of one of the filters from the filter list, the results will be directly shown in the preview window. Below the middle part of the dialog you can change the name of the output layer that will be generated when clicking "Apply" or "Close".

On the right side of the dialog you can preview the outcome of the filters (3). (See documentation on "Preview tool"). In the lower right part you can select whether you want to display the filters over the selected layer or save the filtered image as a new layer (4).

The button "Apply" will apply the changes according to the entered parameters, keeping the Filter dialog open. The "Close" button will apply the changes and close the Filter dialog. The "Cancel" button will close the Filter dialog without applying any filters.

All filters in the filter list can be activated or de-activated through the "Active" checkbox. This checkbox is usually located in the upper part of the filter control panel.

Enhancements

The brightness filter changes the brightness value of the layer. You can increase or decrease the brightness by moving the position of the sliding bar or by entering the value directly in the text box and press enter.

The contrast filter changes the contrast value of the layer. You can increase or decrease the contrast by moving the position of the sliding bar or by entering the value directly in the text box and press enter.

Spatial filters

With this type of filter, graphical transformations like smoothing, edge detection, sharpening etc. are applied to the image.

The following filter types can be applied:

MEDIAN FILTER

The median filter applies a kernel of a certain size, which is determined by the user through the sliding bar labeled Window side.

The median filter is normally used to smoothen and to reduce noise in an image, by moving a kernel of N x N number of pixels over the image and evaluating each central pixel, replacing its value with the median of its neighboring pixels. Compared to the Mean filter, the advantage of the Median filter is that the final pixel value is a value that actually occurs in the image and not an average.

MEAN FILTER

The mean filter applies a kernel of a certain size, which is determined by the user through the sliding bar labeled Window side.

The filter replaces the value of the central pixel with the mean value of the surrounding pixels. Each value of the kernel would be one and the divider would be the total number of elements in the kernel (i.e. a kernel of 3 x 3 would replace the value of the central pixel by the average value of the nine pixels covered by the kernel).

LOW PASS FILTER (smoothing filter)

The low pass filter applies a kernel of a certain size, which is determined by the user through the sliding bar labeled Window side.

Using a low pass filter tends to retain the low frequency information within an image while reducing the high frequency information.

SHARPENING FILTER

By moving the slider to change the sharpness (values from 1-100), the contrast of an image can be changed. The results can be evaluated in the preview window. With a higher contrast, details in the image can be accentuated but the noise will also increase.

GAUSS FILTER

The Gauss filter applies a kernel of a certain size, which is determined by the user through the sliding bar labeled Window side.

The maximum value appears in the central pixel and gradually decreases for pixels that are further away from the central pixel.

CUSTOM FILTER

This is a kernel of 5 x 5 or 3 x 3, for which the values can be introduced by the user. After multiplying the pixel values with the kernel values, the result will be divided by the number specified in the Divisor textbox.

MODE FILTER

The mode filter applies a kernel of a certain size, which is determined by the user through the sliding bar labeled Window side.

This filter takes the value that occurs most in the surrounding pixels and assigns it to the central pixel.

Colour adjustment

Edge detection

These filters attempt, through the use of kernels, to detect edges in the image and change the image so that these edges are enhanced, while the rest of the image is grayed out.

There are four edge detection filters, all with the same interface and options, in which the user chooses a threshold in the range 0-255, and the possibility compare the results by ticking the compare check box:

SOBEL

The Sobel filter detects the horizontal and vertical edges separately on a grayscale image. Colour images are converted to RGB gradations. The result is a transparent image with black lines and some remains of colour.

ROBERTS

The Roberts filter is suitable for detecting diagonal edges. It offers good performance in terms of location. The major drawback of this filter is its extreme sensitivity to noise and therefore has poor detection qualities.

PREWITT

The Prewit filter detects edges in all directions as it consists of 8 kernels that are applied over the image pixel by pixel.

FREI-CHEN

The Frei-Chen filter processes the neighbouring pixels as a function of their distance from the pixel that is being evaluated. The result is that edges in all directions are detected.

Masks

Histogram

Description

To launch the histogram dialog window, use the drop-down toolbar selecting the "Raster Layer" button on the left and "Histogram" in the drop-down button on the right. Make sure that the text box that displays the current layer is set to the name of the raster layer for which you want to see the histogram.

The Histogram dialog shows a histogram of the statistical distribution of pixel values in the current view. This information is often useful when you are trying to color balance an image. In the middle of the dialog you will see the graph on which you can right-click to show a context menu with general options for this kind of graphics.

In the upper part of the dialog (1) are the controls to configure the histogram:

Preferences

The Raster section of the Preferences dialog contains the option "Number of classes" where you can set the number of intervals in which the histogram is divided when the data type of the image is not Byte. For Byte images, this value is 256. In the preferences dialog, the default value of this option is 64 but you can choose any of the options (32, 64, 128, and 256). The intervals are the parts in which the range of values is divided. For example, if we have a DTM with values between 0 and 1 and there are 64 intervals, each interval will have a range of 1/64.

The number of classes does not only refer to histograms but also to other functionalities that require a division in intervals of value ranges.

Radiometric enhancement

The maps that are obtained through digital processing of satellite imagery are useful not only for thematic mapping, but also as a backdrop on which map features can be overlaid. If the visible bands are displayed in a colour composition through the colouring of each band with the corresponding colour gun, it is important that the bands are sufficiently enhanced so that the colours appear more natural. The final display colour depends not only on the direct result of the chosen colour composition but also on the radiometric post-processing. The satellite image map will be more useful as backdrop if the bands are enhanced and displayed in colours that match the natural colours as the human eye perceives them. gvSIG provides the enhancement tools to adjust the colours for each band.

In the following sections the different parts of the dialog are described:

Save View as image

The tool for exporting the view as an image can be accessed from the drop-down toolbar by selecting "Export to raster" on the left button and "Save view to georeferenced raster" on the right button. Make sure that the name of the raster layer that you want to export is set as the current layer in the text box.

A message will appear to inform that you can use the selection tool to set the area in the view to export.

Now, you can select two points in the view to define the rectangle of the area to be exported, by clicking the first point and dragging the mouse towards the second point, then release.

Then, the Save view to georeferenced raster dialog will appear. If the selected area is too small, the dialog will not appear and a bigger rectangle must be selected.

The upper part of the Save view to georeferenced raster dialog shows the coordinates of the two points that define the selected area in the view. You can edit the coordinates to change the selected area.

In the option box in the central part of the dialog you can choose from three selection methods:

• Scale. Selecting this option will enable the Scale textbox and the pull-down box "Spatial resolution" which refers to the resolution in points per pixel (ppp) of the exported image. When entering a value in the Scale textbox and clicking enter, the values "Mts/pixel" and the size (Width and Height) will be recalculated for the output image.
• Mts/pixel (Meters per pixel): When selecting this option, the "Mts/Pixel" textbox is enabled. When you enter a value in the Mts/Pixel textbox and press enter, the values for "Scale" and size ("Width" and "Height") will be recalculated automatically for the output image.
• Size: When selecting this option, the text boxes to enter the "Width" and "Height" will be enabled. When you enter one of these values, the other will be calculated automatically to preserve the right proportions of the image. The other data ("Mtx/Pixel" and "Scale") will also be recalculated automatically. By default, the Width and Height values are displayed in Pixels, but you can select the units (Pixels, Cms, Mms, Mts, or Inches) in which you want to see these values.

NOTE: To save time and memory the maximum size of output images is limited to 20000 x 20000 pixels. If the intended output image is larger and you click on "Apply", gvSIG will display a message that the parameters must be changed before trying again.

Clicking the "Selection" button will open a file browser dialog where you can specify the output file. Depending on the type of file, the corresponding driver will be loaded (you will notice that the button on the right of the "Selection" button will change). For example, an output file .jp2 will open the properties dialog for Jpeg2000. The formats in which you can save are .TIF, .IMG, .BMP, .PGM, .PPM, .MPL, .RST, .JP2, .JPG, and .PNG. Furthermore but only on Linux kernel 2.4 you can also select ECW.

When you select the output file, the Properties button will be enabled.

For example, for geoTiff the dialog will look like this:

• Photometric: [MINISBLACK | MINISWHITE | RGB | CMYK | YCBCR | CIELAB | ICCLAB | ITULAB]. This assigns the photometric interpretation. The default is RGB, as the input image consists of 3 bands of the Byte data type.
• Interleave: [BAND | PIXEL]. By default, tiff files are interleaved by band. Some applications only support interleaved by pixel, in which case you can change this option.
• Compression: [LZW | PACKBITS | DEFLATE | NONE] This refers to the data compression. The default option is NONE.

When the output image is selected and the properties set, you can click on "Apply". A progress bar will appear. Depending on the size of the output file, this process may take while. Processing times may vary between a few seconds or several days, so it is important to check the size of the output image in pixels before clicking "Apply". When finished, a screen with statistics will appear that indicates the path of the output image, the disk size, the duration of the process and whether it was compressed. To check the georeferencing of the output image, you can add it to the view as a new layer with transparency.

Clipping layers

(The clipping tool can be accessed from the raster toolbar by selecting "export to raster" from the left drop-down button and "Clipping" from the drop-down button on the right. Make sure that the raster layer that you want to clip is set as the current layer in the text box.)

With the clipping tool, you can create new layers from an existing one. The options are:

• Extract an area from the input image to be saved as a new layer (cropping)
• Modify the resolution through various interpolation methods
• Modify the order or the number of bands
• Separate the bands into multiple files

Zoom to raster resolution

You can zoom to raster resolution by right-clicking on the layer in the TOC. In the context menu that appears, click "Zoom to raster resolution".

This will activate a crosshair cursor in the gvSIG view which allows users to perform an action by clicking somewhere in the view. The action in this case is that with every mouse-click, the view will be centered on the point where you clicked. In addition, the view will zoom so that one screen pixel is the same size as a pixel in the current raster layer.

Automatic vectorization

The "automatic vectorization" function can be launched from the raster toolbar by selecting "Raster process" on the left drop-down button and "Vectorization" on the drop-down button on the right. Make sure that the name of the raster layer that you want to vectorize is displayed as current layer in the text box.

With automatic vectorization, you can generate a vector layer from a raster image using preprocessing to highlight the features of interest.

When launching the Vectorization dialog, the first step is to select the area of the image that you want to vectorize. Keep in mind that the vectorization process may take a long time, so it is recommended to minimize the area (number of pixels) for vectorization. The selection of the area for vectorization can be done in several ways. You can type the coordinates directly; either in pixel coordinates or in the map coordinates. The area can also be selected from the view by clicking the button "Select from the view", after which you can draw an approximate rectangle to define the area. Another selection option is by Region Of Interest (ROI). You can define a ROI here or use a previously defined ROI to set the area for vectorization. In the section "ROI selection" appears a list of available ROI and a checkbox next to each of these to select one or more ROI that you want to use. There are two options to vectorize the ROI: to vectorize the entire area inside the rectangle (bounding box) that covers all the selected ROI, or vectorize only the areas inside the ROI while considering the values outside the ROI as NoData values, excluding them from the calculations.

Finally you can select the scale of the image to preprocess. This is useful because a higher resolution of the preprocessed image will result in a higher precision for the resulting vector layer. You can define this with the drop-down text box labelled "Output Scale". By default, the resolution will be the same as the input image.

When moving on to the next step of the wizard, the process of cutting the image for preprocessing is started. A progress bar appears with the warning that this operation could take a few minutes. The resulting image cut is saved in the temporary folder of gvSIG.

There are two methods to preprocess a raster image to vectorize. The first is by creating a limited number of grayscale levels from the original image. The image will be converted to grayscale using one single band or a combination of bands (use the drop-down button labelled "Bands"). For the conversion to grayscale, a posterization process is used to reduce the number of different values. (By default, the image is reduced into 2 levels only: black and white.) For this process you can control the threshold on which the values are passing from black to white and vice versa. This can be done by moving the "Treshold" slider while you can see a preview of the result. (The Treshold slider is only available when there are 2 levels; when there are intermediate grayscale levels, the slider is disabled.) In addition to the posterization threshold, you can apply a mode filter or a noise filter to smoothen the result.

The second preprocessing method is useful to vectorize contour lines and can be applied to data types other than byte. With this method you can define intervals between each contour line to be vectorized. You can specify the number of intervals in which you want to divide the raster, or indicate the size of each interval. The cuts that have been selected will be shown on a graph that represents the histogram of the image. On this graph, you can modify the distance between cuts, or add or remove some of them using the mouse. It is also possible to modify the distance between cuts in numeric format using the table on the right of the histogram. Each entry in the table represents a cut with the corresponding value. This type of preprocessing is used for digital elevation models (for example .adf or .asc images).

When moving on to the last step of the vectorization wizard, the preprocessed image is generated with the specified values, and saved in the temporary directory of gvSIG.

The last step is to select the method for generation of vectors. There are two methods: contour and potrace, that can be selected from the drop-down button after which a panel appears with settings that are specific for the method. The first method is the simplest and does not have any options. This method will trace the vectors in straight sections going through the pixel centers. This generates a network of vectors based on very small straight sections. The potrace method uses the potrace library for vectorization. The available options for this method are those that the potrace library provides and they are used to define the precision of the tracing of the curves: number of points for each curve, threshold, optimization, etc.

When clicking on "Apply" or "Accept", the process of vectorization will start after which you will be prompted whether to display the generated layer in the TOC.

Analysis view

The "Analysis view" can be launched from the raster toolbar by selecting "Raster layer" from the left drop-down button and "Analysis View" on the drop-down button on the right. Make sure that the name of the raster layer that you want to analyze is displayed as current layer in the text box.

With this functionality you can zoom in on the current raster layer with 3 different zoom levels:

1. On the left of the view, the layer is added to the locator map of gvSIG. This provides a general view of the layer, and you can zoom into the locator map by clicking and dragging, thus drawing a red rectangle. The area inside the red rectangle will be displayed in the view.
2. The view itself is the second zoom level which functions independently, and the zoom variations that are performed on the view will be applied to the locator map as well so that it keeps being centered on the correct area.
3. When launching the Analysis view, a small floating window labelled "Cursor Zoom" appears in the upper right corner of gvSIG. This window has the highest zoom level. The zoom level is fixed and always centered on the mouse point. By moving the mouse over the gvSIG view, you will see the contents change.

You can change the relation between the zoom level of this floating window and the gvSIG view. This is done by right-clicking on the floating window and selecting one of the values that are shown in the drop-down menu that appears. The available options are x4, x8, x16, and x32. This means that the pixels in the floating window will be 4, 8, 16, or 32 times bigger than the original.

The floating window also shows the RGB values of the pixel on which the cursor is currently located. The text colour of the RGB values as well as the colour of the central cross (red by default) can be changed by right-clicking on the floating window and choosing the option from the drop-down menu.

Keep in mind that, to see the effects in the floating window while moving the mouse over the view, the view must be active. If it is not active, just click on the view. When the cursor is outside the view, the content of the floating window appears black.

There can only be one Analysis view open at any time in gvSIG. Therefore, the button "Analysis View" is re-labelled as "Close Analysis View" when the Analysis view is already open, so that it can be closed before re-opening.