Page 219 - Green - Maritime Archaeology: A Technical Handbook. 2nd ed

198 Maritime Archaeology: A Technical Handbook, Second Edition
process can rapidly get out of hand. A directory structure was established using two separate folders, one for the images and the other for the Pho- toModeler files. By keeping the images separate, they were easy to dupli- cate and to ensure they were not modified (resaving modified JPEG files results in loss of quality as it is a “lossy” compression). Images were kept in folders and named according to the date and time of dive (for example: 010718AM, 010718PM, 010719AM, etc.). With the digital camera the indi- vidual files are date and chronologically named. The PhotoModeler folders were managed in the same way. The targets were named in PhotoModeler with the amphora number (or target number) together with target descrip- tor, L (left), R (right), C (center) for the mapping labels, and T (toe, for the toe of an amphora).
Another problem with Photomodeler (discussed earlier) was when trying to measure very small single artifacts, items otherwise hidden from the camera, and upside down amphora with only the small toe visible for mapping. In the first two seasons these situations were dealt with by taking direct measurements, whereas in the third season with the development of the control grid, PhotoModeler was used.
B. ACCURACY
Estimating the accuracy of any underwater surveying technique is noto- riously difficult. Comparison between the conventional surveying technique and the PhotoModeler system proved interesting (see also discussion in Chapter 4, Section VI.C). The control points were surveyed using tape mea- sures on a number of occasions resulting in a large number of measure- ments. Site Surveyor was then used to process these data, which then provided statistical information on the accuracy of the survey (See Table I). These results represent the best possible accuracy achievable on the site using conventional tape measurements. Given the nature of the control points (some robust others more fragile), it is likely that the points them- selves have an uncertainty in their position which may affect the accuracy. As described previously this gave an average residual of 6 mm with an RMS residual of 9 mm. This was not the accuracy of the survey of the amphora, because the control point calculation was the result of hundreds of mea- surements. It is likely that the manual measurement of an amphora con- sisted of four or five individual measurements for each point and was probably an order of magnitude larger.
PhotoModeler was also used to survey the control points and the results are shown in Table II. By fixing the three coordinate defining points (zero, X, and A) in PhotoModeler the program provided the true coordinates of