1. System
The registration system is a full featured application intended for beginners and advanced users. Registration may be achieved by one simple click or may be controlled by several parameters. A statistical procedure is used to characterize good and bad registrations. Based on this “good fit-bad fit” statistical test the user can stop or modify the parameters or continue the processing.
Implemented tie points extraction algorithms: (1) optical flow and geometry; (2) wavelet transform and (3) closed and open contours (LoG).
Accepted image file formats: BMP, GIF, JPEG, JPEG2000, PNG, RAW, TIFF, GeoTIFF, GeoJP2™, etc.
Georeferencing: Compete mosaic georeferencing through base image, accepting GeoTIFF and GeoJP2™.
Operational systems: Windows (32 bits), UNIX (Linux, Solaris, etc.)
The system has been implemented using common C++ and Trolltech Qt
library which provides platform independence (Solaris, Linux and
Windows).
A universal viewer was developed to simplify image viewing and
handling. It provides basic read/write and preview operations such as:
load file, save file, change zoom, image information, full screen
preview, clipboard handling etc. The visualization module also contains
a layer to handle special information like control points, windows
defined by cursor etc.
2. Toolboxes
System contains toolboxes that increase the registration strength when it gets inputs from the user. There is an image pre-processing module that allows the user to change image resolution, to select a specific band, to enhance the image histogram, etc.
Control points can be displayed and edited in a powerful embedded editor or exported to external applications. Editor visualizes useful information like: differences between points, mean squared error of all or selected points. Selected points may be showed on the images.
3. Registration
Registration can be performed by using small portions (windows) of the images instead of using the whole images. This is useful in cases of cloudy images, multi-temporal ocean shots, etc.
Mosaic procedure contains several useful features like: mosaic both images into separate channels to verify the matching accuracy, equalize the images using simple mean and standard deviation in the overlapping area etc.
Copyright ® 2001-2002,
Image Processing Division, National Institute for Space Research
(INPE), Brazil.
Copyright ® 2001-2002, Vision Lab, ECE, University of California Santa Barbara. |