uSMART Softcopy (DPW)


Improve your turn-around time by using automatic pixel correlation (pixel matching) to create DEMs, assist with aerial triangulation, etc.

Image correlation (also referred to as image/pixel matching) is a process that involves a comparison of the pixels for defined areas of two images. It enables the position of a particular configuration of pixels on one image (the "reference image") to be searched for and found in an area (the "search image") on another image. We now have known X, Y positions of a pixel in the reference image and the search image. From these two sets of image coordinates we can determine a Z value (ground height).

The correlation routines also utilise image overviews which makes the routines much faster. In the grid correlation routine it causes a type of averaging which prevents points from being correlated on isolated trees. It will also prevent points around trees from being "lifted" into the air because of the height of the trees i.e. there will be fewer points that are not on the ground. Areas where there are buildings and clusters of trees will be skipped because of the height tolerance restriction. If a DSM (Digital Surface Model) is required, the user can enter a bigger height tolerance thus allowing the routine to correlate on the buildings, etc

The uSMART Softcopy system utilises automatic pixel correlation in the following routines:

Dynamic Correlation
"Keeps" the floating mark on the ground thus making it possible to map most features automatically without the (up until now) basic photogrammetric requirement of stereo vision.
Relative Correlation

Places points in user specified positions on images forming stereo models and/or connecting models in a strip. These points are then used to determine the relative orientations of images.

Strip Correlation

Correlates points that are common to neighbouring strips used to tie the strips together for aerial triangulation.

Grid Correlation

Place a user-defined grid of points over an area to create a DTM. User defined parameters enable users to control the creation of the grid e.g. to stop the routine from correlating on the tops of trees and buildings without human intervention as a point will be re-calculated until it falls within specification (e.g. height tolerance) or not placed. There are various editing routines that may be used in post processing to "clean" the data i.e. remove high/low points, smooth points, etc.

The user is always in control of the Grid Correlation and the amount of editing that is required. However all these routines can be combined in a batch process to run at once resulting in a smooth regular DTM across the entire job.

Inner Correlation
Determines your inner orientation of scanned images. The user only observes one inner orientation, which is used as a reference when correlating the remaining images. (Not used for digital imagery.)
Calib Correlation
Measures control points on an image necessary for calibrating a camera lens, scanner, etc.