• OrthoEngine 3D Viewing and Editing - 3D Viewing, feature extraction and editing
• OrthoEngine IKONOS Rectification – Orthorectification of Space Imaging’s Ikonos satellite imagery at 1meter and 4 meter resolutions
• OrthoEngine Automatic DEM - Autocorrelation of optical stereo satellite imagery to produce DEMs
• OrthoEngine Automatic RADARSAT DEM – Orthorectification and Stereo DEM production

• Landsat 4, 5 and 7
• SPOT 1,2,3, and 4
• IRS 1-A, 1-B, 1-C, and1-D
• AVHRR (Canadian format only)
• ASTER
• JERS 1
• ERS 1&2
• RADARSAT (all beam modes)

OrthoEngine includes access to PCI Geomatics’ Generic Database Technology (GDB). This allows direct access (without translation) to more than 80 raster and vector formats in their native format. This prevents lengthy import/export operations, making the processing much faster.

Project Setup
Project setup requires the user to specify project, name, input and output projections, formats (i.e. what projection the GCPs are being collected in), whether aerial or satellite modeling, type of sensor, etc.) Large projects may require setting up multiple project files so more than one person can work on the project at one time. This then requires that you export control points between project files if common control points are desired. Project setup is straightforward and well documented. 

Tie Point Collection
When orthorectifying multiple scenes within a project area, tie points are typically acquired. Tie points are produced by identifying the same pixel location in two or more images within their overlap areas. OrthoEngine can process tie points for an unlimited number of input images. Unlike GCPs, tie points do not have georeferencing information. They serve to establish control between images, and are therefore not required when processing a single scene. OrthoEngine has the ability to automatically identify tie points between two or more images. For tie points, root mean square (RMS) residuals are calculated that identify the relative accuracy of each point. Quality control should be performed on auto tie points. 

The bundle adjustment can be performed in OrthoEngine as an iterative process during the GCP and/or tie point collection. This enables the user to determine the accuracy of the block adjustment while in process, rather than at the end. The amount of time required to perform the bundle adjustment is dependent on the number of GCPs collected at that time and the number of images loaded into the project. Alternatively, to save time, the bundle adjustment can be performed at the end of the GCP or tie point collection. 

Residual Reports
To get an overview of the control point residuals and identify possible outliers, Ortho Engine provides the ability to generate residual reports for all scenes within a project or for individual scenes. Outlier points can be easily identified in the table and edited through the use of the Edit Point function. Once the outlier has been identified, this function will load the images of interest, zoom to the highlighted point and enable the GCP Collection panel for editing. 

Orthorectification
The process of orthorectification typically requires correction of terrain displacement through use of a Digital Elevation Model (DEM). Contained within OrthoEngine is a suite of tools providing the capabilities of importing and editing DEMs, if necessary. The software currently supports raster formats including USGS DEM and NIMA DTED. Tools are also provided to generate raster data from vector/point data such as contours, elevation points, break lines and Triangular irregular networks (TINs). OrthoEngine also enables the input GCPs and DEMs to be in different input projections and the output imagery to be in a third projection. This eliminates additional processing steps typically required at the beginning of the project.
  OrthoEngine resampling options include 8 point and 16 point sin x/x, Nearest Neighbor, Bilinear Interpolation, and Cubic Convolution. The sin x/x resampling algorithms are uncommon among commercial image processing software packages. Similar to cubic convolution, they use a larger kernel size (64 pixels for 8 point and 256 pixels for 16 point) and produce slightly better results. However, the trade off is that these resampling methods require more processing time. The software processing speed can also be optimized by adjusting the computers cache allocation. The general rule here is to select a cache size one-half the size of your computer cache (when using Windows NT). Modifying the sampling interval will also affect the processing time. Large projects containing multiple scenes can easily be batch processed to take advantage of off-hours processing. 

Mosaicking Tools
Mosaicking is the process of merging two or more orthorectified images together into a single scene. Mosaicking requires delineation of cutlines and, in many cases, radiometric adjustment of adjacent images to hide the seamlines and produce a more visually pleasing product. OrthoEngine has both manual and automatic mosaicking options. The OrthoEngine automatic mosaicking option is ideal for projects where the following conditions exist:

GCPs can also be imported from a PCI chip database. The Chip Manager module (included with OrthoEngine) stores a small digital image file identifying the location of a single GCP and information about the sensor. It enables use of selected GCPs for multiple projects independent of imagery type and can be queried for criteria including area of interest, sensor type and chip resolution. 

Identified points can be targeted as either GCPs or check points. Independent check points are commonly used to evaluate the overall accuracy of the bundle block adjustment. Stereo points help reduce the total number of points and improve accuracy. Stereo points are GCPs identified in the overlap areas between two or more images. Another feature of the software is that once a model has been established (a model requires at least four GCPs to be collected), the software will automatically estimate the approximate locations for new GCPs on the uncorrected imagery. If satellite ephemeris data is used, approximate locations can be determined with less than four points. Typically the more GCPs collected, the better the software is at identifying new locations. Still, the position is only approximated and quality control and manual position of the point to its actual location is often required.

• radiometry is fairly consistent between images,
• imagery is cloud and haze free,
• there are not significant water bodies, and
• there is not significant geometric mis-registration between scenes. 

OrthoEngine has a complete suite of tools for manually mosaicking imagery. Although more labor intensive, manual cut-lines and radiometric adjustment give the user a lot more control in producing a mosaic that is seamless in appearance. 
Advantages include:

• Defining custom blend widths to adjust the feathering distance between scenes,
• Color balancing of specific scenes to reduce radiometric differences,
• Manual adjustment of brightness,
• Manual delineation of cut-lines based on clouds, terrain variation, roads, and other factors.

Technical Support / Customer Service
PCI has a web-based Customer Service site for receiving questions and providing answers regarding their software. Within North America, they also offer technical service and support via the telephone Monday through Friday from 8:00 am to 6:00 pm EST. Technical support in Europe is available for 15 hours per day from 9:00am to midnight (Greenwich Mean Time). To help provide more specialized support, PCI also has separate web addresses and phone numbers to support software licensing issues. Their new Service Liaison provides support regarding non-technical questions such as status of a software order, and customer feedback. Over the past several years, the authors have received a very good customer service response from PCI with initial response to most requests within 24 hours. 

Documentation/On-line Help 
Textual documentation is provided for all of the PCI OrthoEngine modules. The manuals are well written, comprehensive and easy to reference. They include project tutorials and step by step instructions that walk the user through each phase of the process. Tips are also provided for various aspects of some processes such as GPS and tie point   inexperienced users could benefit from additional fundamental information on photogrammetry concepts. Comprehensive on-line help is included for each module and enables quick search and query abilities. 

Conclusions
Our experience has found that the PCI OrthoEngine Core and Sensor Rectification modules are comprehensive, intuitive and relatively easy to use software ideal for small to medium sized orthorectification projects. The software can be used for larger projects by splitting the project area into multiple files.
  Strong points include OrthoEngine’s flexibility for input formats, and the variety of methods for GCP collection and generating DEMs. Further flexibility is afforded by OrthoEngine’s ability to be used in combination with other orthorectification software packages. For example, OrthoEngine can be used to collect GCPs, while a different software can perform the orthorectification and/or mosaicking. Independently, OrthoEngine offers a complete beginning to end solution for efficient production of highly accurate orthorectified imagery products.