Scanners and software can turn dirty drawings into usable data
Los Angeles, Calif.
Scanners produce files in raster format after they read paper drawings. It’s about the simplest format because it holds just pixel or dot information. CAD systems, on the other hand, produce geometry and features defined by equations and constraints, and store them in what’s called vector data. Vector is generally more useful in downstream applications such as manufacturing. However, raster information is as useful as vector data for many applications thanks to recent software developments.
Early thinking in drawing management was to immediately translate raster data to vector. This is no longer necessary because advances in desktop computers provide new tools, significant savings, and productivity gains to those who need to bring paper drawings directly into an electronic CAD database.
In the past, there were three choices for dealing with raster images: paint programs or low-level tracing methods, hybrid files that stored raster and vector information together, and raster-to-vector converters. Software technology now delivers better tools than simple paint packages for editing raster images, better implementation of raster-vector hybrids, and constraint managers that improve raster-to-vector results.
Surprisingly, several raster editing features such as those that straighten out lines that were tilted during scanning and remove speckles on old drawings, are found in midrange CAD but not in more expensive systems. Higher-priced CAD systems focus on solid modeling and ignore the problem of reading legacy data in other formats. But 10 or 20 years down the road, someone will need to unlock the data in your archive, and raster technology will be the key. Here are several recent ideas for using the often overlooked raster format and how software is making it more useful.
Treat paper as a last resort for archived information. The primary justification for producing raster images is to convert archived paper drawings into more usable digital formats. This “digital paper” can become the release document for reference and distribution. When someone needs just a brief look at a drawing to extract information, no conversion is necessary. A drawing’s raster file can be more quickly located and viewed on screen than hunting for the original paper document.
Raster will be more durable than other formats in the long run. Raster data provides a reliable format for transferring information between different CAD systems, along with Internet viewing and collaborating. Industry observers have remarked that we will be able to view raster files in 50 years. But data stored in a proprietary drawing format will need a conversion when an old copy of the CAD package is not on hand.
Working with raster files takes little training thanks to the introduction of improved hybrid raster-vector editing tools. These make it practical to work with raster data as an active production CAD model.
Many software vendors provide simple editing tools. Simple tools clean up a drawing after it’s scanned. For example, old or original prints may have fingerprints, smudges, marks, and tears that appear as speckles in a scanned image. Paper can shrink or stretch over time, and as part of the scanning process. Drawings can become misaligned when scanned. Additional problems are noticeable when one zooms in on a raster image only to see stair steps in what should be a straight line. Fortunately, software tools such as deskew, line smoothing, despeckle, and rubber sheeting are available to repair these problems.
Basic raster-editing functions in at least one midrange CAD program includes cut, paste, crop, invert, resize, and rotate. These can operate on an entire image or a small area. Users can apply cutand- paste operations within a raster image or even through the Window’s clip board with other applications.
CAD systems with raster features also allow heads-up digitizing, which means users must trace over the raster image with vector elements. While the functions provide considerable power for editing raster images, they are insufficient for modifying mechanical engineering drawings. Consequently, rasterediting systems for engineers provide functions to add geometry such as lines and arcs using CAD techniques. CAD systems are also smart enough to recognize raster entities so that geometry creation commands can snap to end points and intersection points.
Hybrid systems let raster and vector data coexist on the same drawing. These systems either directly, or through optional modules, support raster editing and manipulation. The two data types can overlay one another, but separate editing commands generally work on either raster data or vector data, not both.
The most advanced hybrid raster-vector systems even let users edit raster geometry such as filets, chamfers, trim, and offset. Such systems generate dimension and text entities as well as line styles, and crosshatching. The CAD commands for translate, mirror, erase, relimit, and offset operate on both raster and vector elements.
The smartest CAD programs recognize geometry in raster forms. A major software feature in Helix Drafting recognizes and manipulates raster entities such as lines and circles as if they were vector entities. These raster entities can be deleted, moved, rotated and copied without leaving holes in the raster image. The holes crop up when users remove an entity that crosses another. The better systems recognize holes and fill them in. Without this capability, making even minor CAD edits to a raster image becomes tedious work.
Methods for converting raster to vector are improving. Continued advances in computers are providing the computation horsepower necessary for significant improvements in raster manipulation and accuracy. These capabilities operate on individual entities or groups within specified areas. The challenge is to convert the raster data efficiently into a form that can be edited in a CAD system. This means dashed lines and circles should convert to single entities rather than many short line segments. Failure to produce lines and circles limits the usefulness of the geometry.
Even successful conversions do not eliminate accuracy issues. For example, a draftsman with a straight edge cannot draw a line with the same precision as a CAD system and the intended dimensions are only in the dimension text, not the actual distance between lines, and converting the information (from raster to vector) does not increase its accuracy. For example, drawings that have been scaled improperly are not corrected when converted to vector formats. Also, scanning a good drawing at 100 dpi produces vector information good to about 0.01 in. Manufacturing requirements for 0.005-in. accuracy needs interpretation. Several approaches can deal with the situation.
For one, the operator can treat the drawing as out of scale and manually enter nominal dimension text values. Operators may also interactively recreate geometry in the area of interest using converted raster data as reference. And thirdly, operators can use parametric capabilities in the CAD system to rectify the geometry and adjust it to match the dimension text. Then the geometry can be used as basis for future 2D additions and modifications or as profile data for modeling a solid.