Other Free Encyclopedias » Online Encyclopedia » Encyclopedia - Featured Articles » Contributed Topics from F-J

Image Manipulation: Science Fact or Fiction - Suggested General Guidelines for Scientific Imaging Applications

data enhancement images picture

DAVID MALIN
Anglo-Australian Observatory, RMIT University

In many branches of science, an image can be the primary source of data and the results of an experiment. The integrity of the imaging data may thus be central to a new result or discovery. However, it must also be recognized that no image is totally objective. The making of a picture for science requires an imaging opportunity, the selection of imaging tools, the processing of the data, and the analysis or interpretation of the resultant picture itself. The imaging scientist must choose the moment, the method, the medium, and much else when making an image. At each phase of the input process, subjective choices are inevitable. For this reason, accurate recordkeeping and the preservation of records and original materials has always been an important part of scientific photography.

In an era of digital photography, new opportunities for subjectivity arise in the image processing and output stages. Readily available image editing software encourages the creators of scientific images to make them look as convincing and attractive as possible. This enhancement may be especially tempting when the images are intended for publication. Using the term enhancement in this context may be thought to detract from the integrity of the photograph as a scientific record, and at worst may be construed as a fraudulent. This is not an ethical issue for scientists alone; it concerns anyone who uses images for information or communication.

Changing how an image looks is not a new issue in science. Creative cropping, selective shading, and other forms of enhancement have been practiced in the darkroom from the earliest times, although the skilled darkroom operator was often not the person who made the original picture. In the digital age, however, almost every scientist is his own photographer and imaging expert, hence in “complete” control of input, processing, and output. Many varied and subtle image enhancement procedures can be easily applied, with more or less skill, by anyone with the appropriate software. Often, sharpening, artifact removal, or color enhancement may be done innocently or for purely cosmetic reasons. However, it is all too easy to make an picture more convincing than the data might warrant. It is also nearly impossible to detect, especially in the absence of the original, unaltered image file.

Suggested General Guidelines for Scientific Imaging Applications

Because of the diversity of scientific imaging and the enormous variety of issues associated with them, it is impossible to develop specific guidelines for all disciplines; however, some general principles may apply to all situations. The main issue is that digital images should be regarded as quantitative data from the moment of capture and treated as such.

Any process that can be used to alter the captured data before, during, or after original capture should be recorded, especially if the final image is intended for research or publication, irrespective of the capture medium. Image transmission and reproduction should also be considered as part of this process. In this context, it is imperative for scientific photographers to be alert to the influences and limitations of the recording media, equipment, and techniques, and their effect on the end results.

These effects can, in principle, be reproduced and thus quantified or calibrated if necessary. For an image to be truly considered a scientific record, information about the imaging equipment, the environment, and the detector must be recorded as diligently as experimental technique allows, so that an independent investigator can, if necessary, reproduce the conditions under which the picture was made.

All RAW or equivalent image files used for, or reported in, a publication must be archived in an unaltered state along with the other experimental data, for that is what the native image data was. Lossy image compression (e.g., JPG) always results in some corruption of data and introduction of digital artifacts. For this reason, it is unsuitable for archiving. Re-sizing an image file, irrespective of the file format, also introduces artifacts. If any of the published images are pertinent to the reported results, the native image file(s) should be available to the referees or editors.

If the data is enhanced in some way, that enhancement should be limited to global changes of brightness, contrast, or color balance, and the changes applied should be described in the image caption. If cleaning, re-touching, or sharpening has been applied and the image is central to the result, this clarification of the data should be conspicuously referred to in the text and/or caption. In general, it is not acceptable to distort or move components in an image or to include parts of several images unless for comparison purposes, and then only with clear acknowledgement. Where necessary, grayscales, scale bars, and orientation marks should be included, always without obscuring essential information.

In appropriate cases, the native image files or unaltered copies of them should be available online as a supplement to the publication or on request from the authors. They should be archived with the same diligence as any other raw data.

Image Retrieval - Existing Techniques, Content-Based (CBIR) Systems [next] [back] Image Inpainting

User Comments

Your email address will be altered so spam harvesting bots can't read it easily.
Hide my email completely instead?

Cancel or