From the Ground Up
Fortunately, surveying and engineering professionals now have the desktop computer systems available to handle the relatively large image files that are common with digital orthophotos. In fact, the technology available to us today is nothing short of amazing. I checked with one of the most popular desktop providers just before completing this article. A new top-of-the-line computer outfitted with a 1.7 GHz Pentium 4 processor, 1024 Mb of RAM, a 19-inch monitor, and an 80 Gigabyte (GB) hard drive was less than $3,500. And that’s enough computing power for handling very high-resolution digital image files.
Changes have also taken place on the production side of the equation. For example, the computer hardware and software needed to produce orthos in the mapping professional’s office required an investment of more than a quarter of a million dollars in the mid-1990s. Thankfully the technology has improved, competition among software vendors has intensified, and the cost for both hardware and ortho software has dropped to a small fraction of the cost of just a few short years ago. What does this mean to surveying and engineering professionals using digital orthophotos? Orthophotos are now cost effective for most projects undertaken by today’s professionals.
Why Orthophotos?Raw images have no orientation, precise scale or inherent coordinate system. More importantly, however, raw images have certain distortions in them due to a number of factors. A digital orthophoto has all the effects of these distortions removed, is precisely oriented and scaled, and therefore is a precise image representation of the ground. An orthophoto has accurate coordinates established for all areas within the digital image. You can bring up a digital orthophoto in a CAD or GIS system, point to a feature found in the image, and extract the precise coordinate position of that feature. Likewise, you can measure between two visible features in the orthophoto and be presented with the accurate distance between those two features. In short, an orthophoto can be as accurate as the conventional line based maps that professionals have been working with for decades.
The distortions in a raw image arise from two main sources. First, the effects of any tip or tilt of the camera at the time of photo capture create error in the photograph. Secondly, the effects of relief displacement due to height variation of the earth captured within the photograph distort the raw image. If the ground captured in each photograph is perfectly flat and the camera pointed perfectly perpendicular to the ground, then there is no need for orthorectification. But obviously this is never the case in the real world.
The first source of distortion is fairly easy to understand. The second source, however, is much more difficult for someone outside of the mapping industry to visualize. It also is the most significant reason that orthorectification is required when you want to put a digital image to practical use in your profession.
So what is this concept of relief displacement? As the ground rises up, objects on the ground are shifted outward (from the center of the photograph) from their true position. I’ll use the classic photographic example to illustrate this concept. Figure 1 shows a photograph of a group of cooling towers at a power plant. Obviously the towers at this plant were constructed in a vertical fashion. Therefore the coordinate position of the top of the tower should be equal to the position of the bottom of the tower. But notice in this photograph how each of the towers appears to be leaning. Also notice that all towers are leaning in the same direction.
Orthorectification software uses an elevation model of the ground to accurately predict and remove the effects of relief displacement within the image. The process is fairly rigorous mathematically, but easy to program and apply in the computer environment. Orthorectification software examines every pixel within the digital image, extracts an elevation for that position from the elevation model, determines the amount of image displacement due to the ground elevation, and moves the pixel of information to its correct location. After all pixels have been moved accordingly, the software stitches them back together into a seamless image. Figure 2 is a great illustration of what happens to a raw image as the distortions are removed. The once perfectly square image has made it through ortho processing and clearly shows the effects of the distortion along the now very irregular edges of the image.
Image ResolutionOne of the most important decisions that must be made early in the planning of an ortho project is the selection of a resolution. The resolution (and desired accuracy) will determine the flying height for the aerial photography. Ultimately, the resolution will therefore be a significant factor in the cost of the project, the features that you will be able to recognize in the imagery, and the size of the images that you will be working with in the digital world.
As you zoom in on a digital ortho you will begin to see the individual pixels that make up the image. A pixel (short for picture element) is the most fundamental piece of information found in a digital image. All digital orthophotos have a precisely defined image resolution in ground units. For example, consider an ortho with a ground resolution of 1 foot. Imagine going out on the ground and laying out a precise grid with east-west and north-south lines both spaced precisely 1 foot apart. Each resulting 1-foot square on the ground would be represented in the digital image by one pixel of information illustrating the predominant color of that area on the ground. When the grid overlays a sidewalk, the pixel value would be light in color to reflect the visual characteristics of concrete. Conversely, when the grid overlays a fresh asphalt roadway, the pixel value would be dark in color. There are obviously a lot of color variations in between these two extremes.
File SizeA quick note has to be made concerning file size. Yes, the computers are much faster today with larger hard drives and more RAM. All are very important factors. But file size remains a real issue when dealing with digital orthophotos. Calculating the file size is a fairly simple task of multiplying the width of a digital image in pixels by the height, again in pixels.
For example, consider the number of pixels in an image displayed at a scale of 1" = 100' at a size of 20" x 30" when a ground resolution of 1 foot is selected. It should be apparent that this image covers a ground area of 2,000 by 3,000 feet. Using similar logic, this image would be comprised of a matrix of 2,000 by 3,000 pixels at the 1-foot resolution. Therefore the image would require 6 million bytes of storage, or about 6 megabytes. What if the resolution is increased to a resolution of 0.5 feet? Most would guess the file size would double from 6 to 12 megabytes. But actually the file size will quadruple to 24 megabytes. Think about it… it takes 4 pixels at the 0.5-foot resolution to equal one pixel at the 1-foot resolution. Want color in your imagery? Multiply these values by 3 because each pixel of information now contains 3 bytes of information… one byte for each of the red, green and blue bands.