Long before digital imaging had become feasible, various techniques had been developed to improve the quality of photographic recordings. Because the general goal of photography is to convert light from physical objects into permanent images, methods used to modify the quality of the images are extensions of the basic photographic method itself. Every photograph is dependent on a number of choices by the photographer (which film or filters to use, which exposure, etc.). In that sense, no photograph is objective, and the various ways of improving photographic quality are part and parcel of photography.
The advent of digital imaging has expanded the possibilities of handling images, but has not changed anything in principle with regard to the non-objectivity of photographs. In fact, many techniques built into digital imaging systems are adopted from pre-digital techniques. For example, the much-used unsharp mask, an effective algorithm for increasing the apparent sharpness of digital images, stems from an ingenious nondigital technique (e.g., Malin 1993) by which a deliberately blurred positive copy of a negative is superimposed upon the original negative. Similar areas (i.e., those that are blurred in both original and copy) tend to cancel each other whereas the nonsimilar ones (i.e., sharp in original but blurred in copy) do not. The effect is that sharp areas in the original are enhanced, blurred subdued.
In many areas of science, image-improvement techniques are standard procedures to bring out visual detail in recordings from various types of devices. In fact, one could argue that every recording from an instrument, whether a camera or a seismograph, needs to be improved by various methods in order to be useful.
Clearly, however, digital techniques have also made radical image manipulation very easy, and so the question of the objectivity, or truthfulness, of a photograph has become even more vexing than was the case previously. The issue is now purely ethical, not technical. Manipulating an image with the intention to deceive is fraudulent, equivalent to fudging experimental data. Thus, as with experimental data, any deviation from standard procedure in obtaining an image must be accounted for. The recommendation by Hughes (1999) to include the original image alongside the processed images is well taken.
Based on the purpose behind the procedure, we may distinguish among restoration, enhancement, and manipulation of photographic images.
Restoration is intended to overcome the limitations of the recording device, to bring an image as close as possible to what was originally perceived by the eye, naked or through an instrument using visible light.
Enhancement is intended to bring out certain features of the image, in a way different from what the eye perceives. Examples are false colours, edge accentuation, and retrodeformation.
Manipulation is intended to add information to an image that was not originally in it, for example by retouching, or drawing in of contours.
There is no distinct boundary between restoration and enhancement, but both may be said to employ objective procedures, acting equally upon the whole picture using some predetermined method or algorithm. Manipulation, in this context, implies that different parts of the images are treated differently as a result of ad hoc decisions by the operator.
A specific case is retrodeformation, restoring deformed fossils or sediments (e.g., Hughes 1999). Although such procedures technically amount to deformation of the photographic image, the algorithms employed are typically applied to whole images. Because the specific purpose is to visualize an earlier existing state of the object, this procedure is to be regarded as restoration or enhancement, rather than manipulation.
Techniques for image processing are today generally available with the common access to digital imaging programs, such as Adobe Photoshop, and more specific software for image enhancement and analysis. Restoring or enhancing images beyond the basic adjustment of brightness and contrast is still not common practice. However, the techniques now available to nearly all palaeontologists are very powerful for solving age-old imaging problems, such as how to bring out fossils with no appreciable relief or colour difference from the background.
The possibilities of transforming images using digital techniques are endless. Most applications of these techniques add nothing to the visible information content of the images, but rather amount to distortion. Except for their possible aesthetic value, they are of little interest for scientific imaging. The methods discussed here are intended to restore and enhance, not manipulate, the informative value of pictures of fossils.
A digital camera is not crucial to the techniques described here, because the images can also be digitized with the help of an ordinary flatbed scanner or slide scanner. However, in addition to providing a more direct path between object and picture, digital cameras help experimentation in that the results of each exposure are immediately visible.