Architectural and engineering drawings stored by an organization are often created over many years using multiple systems. However, they can be effectively managed by taking a unified approach to how they are indexed and linked. A unified approach includes a machine-based analysis of:

  • All drawing formats, including scanned paper drawings and native CAD/CAM files.
  • All types of drawings including:
    • As designed
    • As built
    • As modified

By including all formats and types of drawings this approach can show what was in place at any given point in time. With effective indexing and linking, users can do things like:

  • Move from viewing electrical schematics for one floor to the electrical schematic on the floors immediately above or below the one being examined.
  • Browse from the as-designed version of a section of a plant to the as-built version, and then, in chronological order to the as-modified versions.

There are two aspects to indexing or linking drawings: the first is to analyze the information contained on title, notes, and revisions blocks on drawings, and the second is to be able to associate drawings of similar objects regardless of the information in such blocks.

Information in Title, Notes, and Revisions Blocks

Information about the subject of the drawing is often embedded in the title, notes, and revisions blocks. Tagging this information or placing it in specified fields enables users to find desired drawings and the other drawings that are related to them. The key is to have systems that can analyze and extract information from drawings as large as size E architectural or engineering drawings (or even larger) scanned at multiple resolutions, while also being able to interpret and extract information from native CAD/CAM-type software files.

Auto-tagging or placing data from the blocks in correct fields uses at least three techniques:

  • Explicit labeling, e.g., the name of the field and a colon precede the data values or appears just above the data values
  • Relative location within unlabeled blocks
  • Block-specific taxonomies, e.g., terms with certain prefixes refer to earlier drawings when used in revisions blocks

Advanced tagging or attribute extraction technology can can take much of the work of indexing and linking drawings off the table in terms of what must be manually provided.

Associating Visually-Similar Drawings

When users are working with specific drawings, they may need to find where visually-similar drawings have been used or referenced e.g., in presentations or memos. They may also like to find where the drawing has been used at different facilities, plants or for different customers. Visual classification technology provides a way to locate other instances of the drawing or drawings that are substantially similar. This works even if there is no information available from title, notes, or revisions blocks, or if the information in those blocks is different from that contained in the drawing of interest.

Visual similarity technology also enables the tagging of drawings that contain specified elements or glyphs even if they are not standard ASCII character sets, e.g., standard or custom drawing symbols.

Finding Relevant CAD/CAM Versions

CAD/CAM systems are vector-based and are more accurate and easier to work than raster or bit image copies. As an example, when drawings are originally drafted, they have a stated scale. However, several things can make the scales inaccurate on subsequent scanned or printed versions of drawings:

  • Image Capture. When drawings are scanned, scanner feed issues can cause a distortion along the direction of feed, either compressing or stretching parts of the image.
  • Saving to File. When scanned images are saved to a file such as TIF or PDF, they may be shrunken slightly to permit the addition of page numbers or legends outside the area taken by the drawing.
  • Printing. When drawings are printed, they may not be printed on the size paper originally contemplated or the print software may shrink the image slightly to fit within the given print area of particular printers.

Even the most accurate indexing and linking of legacy raster or bitmap drawings won’t convert raster/bitmapped images directly to vector-based files. However, where raster and CAD/CAM native files are co-mingled within an organization, this approach does result in identifying which scanned drawings and which CAD/CAM files relate to the same subject. When there are CAD/CAM files, they can be updated with changes from the as-built, or as-modified drawings to yield current CAD/CAM files.

Bottom Line

This approach provides quick access to the legacy drawings that were used in constructing and maintaining buildings and plants and can help considerably in locating relevant CAD/CAM files. This approach can also be used with many other documents or files that are often considered “over-sized,” e.g., maps, plats, schematics, and well logs.

See also my earlier posting, “Using Glyphs to Deconstruct, Classify, and Attribute Files.” <LINK>


For more information on managing content, download your free personal copy of my recent book, Guide to Managing Unstructured Content at http://beyondrecognition.net/download-john-martins-guide-to-managing-unstructured-content/.

Comments are closed.