17
jimFusion Satellite image manipulation SOFTWARE FEATURES QUICK GUIDE * jimFusion was made almost specifically for research purposes and it does not intend to replace well established SIG or image manipulation software. Also notice that its support is limited to the “crew” available in CERENA research center.

jimFusion_presentation

Embed Size (px)

Citation preview

Page 1: jimFusion_presentation

jimFusion Satellite image manipulation

SOFTWARE FEATURES QUICK GUIDE

* jimFusion was made almost specifically for research purposes and it does not intend to replace well established SIG or image manipulation software. Also notice that its support is limited to the “crew” available in CERENA research center.

Page 2: jimFusion_presentation

Developer tools

Base tools: • Python • Numpy • Scipy

User interface: • wxPython

3D visualization: • Mayavi

Image manipulation: • PIL

Page 3: jimFusion_presentation

Main menu

TOOLBAR

OBJECTS

VIEWER

There are three main areas in this software: toolbar, object management, and object viewer. Everything can be done through these sections*.

* The layout of the software may change depending on the version (on the left is the first version).

Page 4: jimFusion_presentation

Main menu - toolbar

With this button you can import image files (TIFF). It will appear a file dialog from which you can select the file you want to import.

With this button you can import columned ASCII files. It will appear a file dialog from which you can select the file you want to import. The user will have to input the number of blocks in each direction and if more than one column exists one file only the last one will be considered.

From here you can export data that is inside jimFusion into either image (TIFF) or

ASCII columned files. The TIFF uses a type of variable that goes from 0 to 255 so make sure that you normalize your data before you export it.

Here you can make all sorts of definitions in the way you want to view or interpret your data. Other specifications (like background color and so on) can only be done in the viewer.

This is the “about this software” information. Who’s the developer, license, and so on...

Page 5: jimFusion_presentation

Import Data

If you select import ASCII you should give the number of blocks in each direction and than select the file you wish. For TIFF the process is even simpler. Just load the TIFF file directly from the file dialog. After data the object will appear in the object management section.

Page 6: jimFusion_presentation

Main menu – object management

In the object management section this is what you’ll see. Notice that no object exists yet on the manager. But if so I would appear with its name (when you load a file a name will be given to that object but can be changed easily) like a list in the white square.

A single click will get the object selected (up). A double click will also bring the object to the viewer (down).

NOTE: Right click will delete (if user says yes) object.

Page 7: jimFusion_presentation

Main menu – object specifics

For organization purposes you can change some details of your object when selecting it. If you change the name, size or first coordinate in the object specifics section (you must press enter to actually change something) you’ll notice that it will actually change those specifics on the object. The only essential information is the number of blocks in each direction.

Page 8: jimFusion_presentation

Object viewer

The object viewer only views one object at a time but its pretty easy to move around. Its mouse and keyboard interactive (make sure you click on the viewer to manipulate it) and you can edit some of its traits by going to the viewer “configure the scene” button.

By default the background is silver white but I’ve changed it to white (foreground -> labels to black) and now I have something more of my preference.

Page 9: jimFusion_presentation

Object viewer – Types and layout

The buttons on the viewer are pretty self explaining. You can choose the perspective (X,Y,Z, isometric), insert orientation glyph or toggle full screen. You also have a save icon that plots the current figure to file. There are several types of objects the viewer can plot. Such as image plot (left figure), histogram and variogram (below).

Plots such as variogram are dynamic which means there can be changes in real time to adequate a variogram model, for instance. But there’s a catch. Since all happens in an equal scale in all axis the value of the variogram or the probability in histogram are not true values but manipulated values to plot something easily visualized on screen. Interiorly jimFusion works with the real values so the user can

work with what it’s visualizing.

Page 10: jimFusion_presentation

User definitions

• Use variable range: Auto for natural variable range of each object, Full to the 0 to 255 range.

• Use colormap: colormap to plot variables, default is jet (blue to red, passing on green and yellow).

• Show as: Image, Surface or Contour. If Surface the variable range will be changed (surface gives spatial weight to the value of the variable representing a 3D object).

• Division for Zscale: This is the number for each the variable of an object is divided when shown as surface (because 255, for example, will give a pretty high altitude for the surface making it difficult to visualize).

Image Surface

Page 11: jimFusion_presentation

Types of plot

There are three main types of plot objects: • Spatial: double click on the object. • Histogram: press the histogram button. • Variogram: press variogram button.

Image Surface Contour

Spatial

Histogram Variogram

Page 12: jimFusion_presentation

Univariate section

The Univariate section gives you access to both histogram (that we’ve seen before) and variable transformations (changing the variable to range = 0->255). Available variable transformations are: • Equalization: meaning histogram equalization. Good to

improve image contrast. • Normalization: transforms from variable range to 0->255

range with linear interpolation. • Sigmoid: used sigmoid exponential function to transform

variable.

Histogram equalization

Normalization

Sigmoid

Page 13: jimFusion_presentation

Filtering and attributes section

The filtering and attributes section gives us algorithms to create new object through filters and attributes (that’s unexpected!!!)*. The attributes go from autocorrelation (using one or two images) to correlograms. In the filtering part we have three of choice, each with its own purpose. Uniform filter, Median filter and Gaussian filter. For the Uniform and Median, the number on the right means moving window size. For Gaussian means standard deviation.

Uniform

Median

Gaussian

* In the first version of the software there is a lot of research being conducted in the attributes sections and some algorithms may give unexpected results.

Page 14: jimFusion_presentation

Slicing and manipulation section

This sections allows us to crop our objects spatially (for instance in a grid of 100 by 100 blocks, creating a new object from the first 50 blocks in each direction), better said slicing. Also allows us to resize (using nearest criteria in the first version of the software) our object by giving the new number of blocks in each direction.

Resize

Slice

Page 15: jimFusion_presentation

Variography section

The variogram button calculates the experimental variogram of the object (if big may take a while, see the progress bar). After finishing the variogram object will appear in the viewer. You can choose which model to use, the sill (screen values), the range in each direction. For you to see the blue line of the model you need to check the checkbox at the right of the model choice. Once you’ve done it the blue lines should appear and be automatically refreshed every time the sliders and sill values are changed.

Notice that even the variogram is 3D so you’ll be having points over the X axis for variogram in the X direction and over the Y axis for the Y direction.

Page 16: jimFusion_presentation

Downscaling data

With resize algorithm you can actually downscale your grid but this method is far from flawless since the user doesn’t have any control on the downscaling process (it uses nearest criteria*).

* Further versions may have other choices for resize using the PIL library, such as BILINEAR, BICUBIC or ANTIALIAS.

Page 17: jimFusion_presentation

template