Converting paper and scanned drawings to CAD

1 Introduction
2 Identifying the type of drawing
3 Methods of converting drawings to CAD
4 Getting the scale right
5 Checking the converted drawing
6 Related articles on Designing Buildings

[edit] Introduction

Many existing buildings have drawings that are available only as paper documents, including dyeline prints, original tracings, photocopies and scanned PDFs. Before an extension, refurbishment or change of use, these drawings may need to be converted into computer-aided design (CAD) format so that they can be edited, measured and incorporated into new design work.

The appropriate conversion method depends on the condition and format of the original drawing, the required level of accuracy and the intended use of the resulting CAD files. Options range from importing existing vector data to manually redrawing scanned images, automated raster-to-vector conversion and measured building surveys.

[edit] Identifying the type of drawing

The first step is to establish whether the source file contains vector information or raster image data.

A PDF exported directly from a CAD application can contain vector information, including lines, arcs, polylines and text. These elements retain their geometric properties and can often be imported into another CAD drawing as editable objects.

A scanned PDF or photograph of a drawing consists primarily of raster image data, represented by pixels. Although the drawing may look identical to a vector PDF when viewed on screen, its lines and text are not necessarily recognised as individual CAD objects.

This distinction affects the conversion process. CAD software can import vector geometry directly from a suitable PDF, whereas scanned drawings generally need to be traced or processed using raster-to-vector conversion software.

A practical way to distinguish the two formats is to zoom in closely. Vector lines generally remain smooth and sharply defined, whereas raster images eventually reveal individual pixels. However, some PDFs contain a mixture of vector and raster information, so the presence of sharp lines does not necessarily mean that the entire drawing is vector-based.

[edit] Methods of converting drawings to CAD

[edit] Importing vector PDF drawings

Where a PDF contains vector geometry, it may be possible to import the information directly into CAD software. For example, AutoCAD includes the PDFIMPORT command, which can convert supported vector geometry, text and other information into CAD objects.

Depending on the original PDF, the imported information may include lines, arcs, solid fills and text. Some layer and object information may also be retained, although the result will not necessarily reproduce the original CAD structure or object properties.

Imported drawings should be checked for missing elements, fragmented geometry, incorrect text recognition and discrepancies in scale. Where a PDF contains raster images, these may need to be handled separately rather than converted into editable vector objects.

[edit] Manual redrawing

Manual redrawing involves inserting the scanned drawing into a CAD file as a reference image, establishing its scale and tracing the relevant information using CAD drawing tools.

This approach allows the drafter to interpret the original drawing rather than reproduce every mark mechanically. Unclear details, damaged areas and inconsistent line work can be assessed, while the resulting geometry can be organised into appropriate layers and drawing conventions.

Manual redrawing can produce clean, editable CAD drawings, but it is relatively labour-intensive. The time required depends on the size, complexity and condition of the source drawing, as well as the level of detail required.

[edit] Raster-to-vector conversion

Raster-to-vector software analyses pixels in a scanned drawing and attempts to convert lines, shapes and text into vector geometry.

The method can be effective for relatively simple drawings with clear, consistent line work, such as site boundaries, component profiles and uncomplicated technical drawings. However, architectural drawings often contain overlapping information, hatching, annotations, handwritten notes, stains and folds. Automated tracing may interpret these as drawing geometry, producing fragmented or irregular outlines that require substantial cleaning.

For example, a wall represented by two parallel lines may be converted into separate polylines rather than a single intelligent wall object. Text may also be converted into outlines instead of editable text. The resulting CAD file therefore requires checking and editing before it can be relied upon.

[edit] Automated and AI-assisted drawing recognition

Automated drawing recognition tools use image-processing techniques, machine learning and artificial intelligence to identify features within scanned plans. Depending on the software, these may include walls, doors, windows, rooms and other architectural elements.

Rather than simply tracing pixels, the software attempts to interpret the drawing and reconstruct recognised features as CAD objects, sometimes assigning them to separate layers or using them to generate a three-dimensional model.

These methods can reduce the time required to convert clear and relatively standardised drawings. However, results depend on image quality, drawing conventions and the complexity of the source material. Dense, damaged or unusual drawings can lead to incorrectly identified or omitted elements. Automated outputs should therefore be checked against the original drawings and corrected before use.

[edit] Measured building surveys

Where existing drawings are known to be inaccurate, incomplete or out of date, converting them directly into CAD may simply reproduce existing errors.

A measured building survey provides an alternative by recording the actual dimensions and arrangement of the building. Measurements can be taken using tape measures, electronic distance meters, laser measuring equipment or three-dimensional laser scanning, depending on the required accuracy and level of detail.

The resulting survey information can be used to produce new CAD drawings or to correct and supplement existing drawings. This approach is particularly important where design decisions, construction work or the fabrication of components depend on accurate existing dimensions.

[edit] Getting the scale right

Scanned drawings do not inherently contain reliable CAD measurement units or physical dimensions. The scale must therefore be established using known dimensions or other reliable reference information.

Where a dimension is shown on the original drawing, it can be used to calibrate the inserted image. In AutoCAD, for example, the SCALE command with the Reference option can be used to resize the image to match a known measurement.

Where possible, scale should be established using a relatively long dimension, as this reduces the proportional effect of measurement and selection errors. A second dimension should then be checked, preferably in a different direction, to identify any distortion.

Scanners and photocopiers can introduce small distortions, including unequal scaling in horizontal and vertical directions. If this occurs, uniform scaling may not be sufficient to achieve dimensional accuracy across the entire drawing. Any correction involving non-uniform scaling should be assessed carefully, as it can distort the original geometry.

Photographs of drawings taken at an angle can also introduce perspective distortion. These should ideally be retaken with the camera perpendicular to the drawing surface. Where this is not possible, image-processing software can be used to correct perspective before the image is imported into CAD.

The original drawing scale should not be assumed to be correct simply because it is stated on the sheet. Reproduction, photocopying and resizing may have altered the dimensions.

[edit] Checking the converted drawing

Regardless of the conversion method, the resulting CAD drawing should be checked against the original before being used for design or construction purposes.

Important checks include:

Overall dimensions and principal room sizes.
Positions and numbers of doors and windows.
Wall thicknesses and the distinction between structural and non-structural elements.
Alignment of walls, openings and other principal features.
Completeness of the drawing, including details that may have been faint or obscured in the original.
Text, annotations, room names and dimensions, including whether these are editable CAD text or simply converted outlines.
Layer organisation and the consistency of drawing conventions.

Particular attention should be paid to wall thicknesses, which can be misrepresented by automated tracing where the software interprets both wall edges as unrelated lines. Dimensions and annotations should also be checked to ensure that they have not been mistaken for physical geometry.

The accuracy of a converted drawing is limited by the quality and accuracy of the original information. A clear, correctly scaled scan does not necessarily represent the actual building accurately, particularly where alterations have taken place since the drawing was produced.

Where design or construction work depends on precise dimensions, critical measurements should be verified on site. Any information that has been measured, inferred, copied or assumed should be identified appropriately so that subsequent users understand the reliability and limitations of the drawing.