GISdevelopment.net ---> AARS ---> ACRS 1990 ---> Digital Image Processing The use of image processing systems for the analysis of digitised Aerial Photography N.C. Coops, D.D. Fraser, N.M. Rollings, G.P.Ellis RMIT Centre for Remote Sensing Victoria University of Technology GPO Box 2476V Melbourne Victoria 3001 Australia Abstract The use of image processing systems in the analysis of remotely sensed data such as that derive from satellites is well documented. There remains, however, a substantial amount of data in the form of aerial photography, analogue satellite imagery and maps, the analysis of which can be greatly improved using image enhancement and geographic manipulation techniques. Aerial photography provides high resolution information and offers an historical archive dating back many years. Mapped data provide useful ancillary information, which can be used to assist in the analysis. Image enhancement techniques may be applied to digital data derived from analogue sources enabling subtle features to be extracted that were not apparent in the original data set. This principle of digitising analogue data has been applied to several data sets relating to geological and land use mapping applications. The use of techniques such as contrast enhancement, histogram equalisation, principal component analysis, decorrelation stretching and classification has allowed significant feature extraction. The digital nature of these data makes them amenable to further analysis in a Geographic information system (GIS). The use of automated cartographic software packages also permits the output of useful and attractive map products from such integrated data sets. Introduction Technological advancement in microcomputer hardware has meant a substantial decrease in their cost with a corresponding increase in their computing power. This has led to an increasing use and awareness of personal computers in their application to tasks previously restricted to expensive mainframes. More recently, the development of low cost scanners (automatic raster digitisers) has emerged as a significant image processing tool for the mapping and graphics industry. Civil and Municipal Engineers GIS specialists and other workers in the spatial data field all hold archives of large-format maps that require regular updating and revision, and from which copies are regularly made. It is only recently that these groups are recognising the advantages of raster data as an archive and raster editing for updating (Bosma et al, 1989). Even with this increase in the use of raster data by a wide variety of users, the purchase of remotely sensed imagery from satellites and airborne scanners at a suitable mapping resolution is still an expensive venture (corblet,1990). As an alternative some users are turning to existing data sources as a means of obtaining high resolution data over particular regions. Aerial photography has been flown in many countries as an integral part of topographic mapping programs. These data, apart from providing topographic information to allow the production of maps, provide a continuous, large photographic scale archive in some cases dating back many years. The digitising of aerial photographs and subsequent entry into an image processing system allows digital manipulation of the data and consequently provides a powerful means of interpretation. The image processing techniques used on digitised analogue images will be discussed in relation to three pilot study areas. Introduction to the study areas Flinders Ranges: The Flinders ranges study area is situated in Southern Australia 75km NNE of port Augusta. The region is highly prospective for mineral deposits which include gold, copper, and industrial minerals. Accurate lithological mapping is paramount to successful ore deposit delineation in this region (Rollings,1985). Existing 1:250,000 geological maps of the area contain insufficient detail with respect to the smaller outcrops which can be important for exploration activities. Very subtle spectral differences exist between many of the rock units (Rollings,1988) Only by using sophisticated image analysis techniques can these subtle differences be exploited for the purpose of lithological mapping . Remotely sensed data derived from satellite sensors lacks the required spatial resolution necessary for structural mapping at a scale appropriate for detecting the ore deposits of the region. Digitised aerial photography algorithms designed to extract line and edge features can be used to enhance small geological structures. For this study area colour aerial photography acquired on the 20/6/84at 1:40,000 was used. Figure 1: Location map of study areas. Warrenbayne: The Warrenbayne study area near the town of Benalla, in Southern Australia, is predominantly grazing land affected by dryland salinity. Groundwater recharge and discharge areas therefore need to be mapped before treatment is undertaken (Fraser et al , 1989) In the past the geographic data of the area has been in analogue form. As a result, detailed interpretation of the data has been limited. Once in digital format the data can undergo more extensive analysis. The aim of digitising the aerial photography was to provide an objective image of the farm areas at high resoultion. This image then be used as a base to display other data, such as thematic classification and annotation (Ellis & Fraser, 1990). The aerial photography for the Warrenbayne area is monochrome, at a scale of 1:25,000 , taken on the 14/12/87. Philip Island : Cowes, the major town on Phillip Island, 120 km SE of Melbourne is a small coastal community relying on fishing and tourism for is economic survival. The Township itself was chosen as a study area for a number of resons: the area itself has a diversity of land use, both urban and rural , with large areas of parkland and nature reserve. the area has been covered by color aerial photography at a scale of 1:10,000. large scale maps are available. The aerial photography for Phillip Island was taken as part of the colour. 1:10,000 coastal mapping series on the 22/10/83. Method of data input Analogue photographs can be digitised into processing system using a variety of techniques. Two every common techniques are: Frame Grabbing: The simplest method producing raster data is by frame grabbing the imagery . This is a low cost option because much of the equipment is readily available. To frame grab an image requires a video camera connected to a personal computer. In this particular application a standard video camera was connected to a special board within the personal computer. This board frame grabs the analogue data and converts it into 8 bit in a 512 by 512 pixel image. This image can then the easily imported into an image analysis system. Electronic Scanning There are a wide range of scanners on the market that are able to digitise aerial photographs, included with many of these scanners are software programs that allow zooming, basic filtering and printing of the imported images. For this particular application, additional software items were not required, since the image could easily be manipulated within the image processing system. The technical specifications of the majority of scanners on the market are similar, varying mainly in accepted sheet size and scanning resolution. For an excellent summary of the types and specifications of image scanners see Bosma et al (1989) and Bosma and Drummond (1989). A flatbed scanner was used which scans at a rate 25 milliseconds per line and provides a maximum resolution of 300 dots per inch (dpi) . Images are scanned in 8 bit or 24 bit resolution using a charge coupled device (CCD) With 8 bit resolution 256 levels of grey can be scanned and displayed in monochrome. With the 24 bit resolution the full 16.8 million colours are available, however, the 24 bit mode requires high resolution video adaptors and monitors on the computer system. (Harrison, 1990). Both black and white (one pass) and colour separations are possible (multiple passes using a set of red, green and blue fluorescent lights). Image processing Data quality: Once the scanned image is in a digital form, image enhancement techniques can be applied to the data. Linear contrast stretches where initially performed. The images exhibited a large dynamic range and so were particularly amenable to contrast enhancement. In order to assess the quality of the data, principal component transformations were carried out. These were computed in the first instance on the Flinders Ranges data (multi spectral). From these the noise to signal ration (NSR) were calculated for both the scanned and the frame grabbed imagery. The Noise to Signal ratio (NSR) is a statistic calculated by the principal component analysis algorithm which summarises the variance of a component relative to the whole data set (Harrision and Jupp, 1990). Typically, the ratio of noise to signal ratio for remotely sensed data should be low at around 5-10 %. Frame Grabbed Imagery: A NSR of 45% was computed for the grabbed imagery. This was expected as the data exhibited a large amount of periodic noise when viewed on the screen. In an attempt to 'clean ' the frame grabbed imagery a median filter was applied to the raw data. This was unsuccessful due to the dominating nature of the periodic noise. Examination of principal component (PC) 2 revealed that this band contained the majority of the noise. Therefore, the median filter was applied to this but was also unsuccessful. A median filter applied to PC 1 produced an acceptable result. A second pass using a median filter removed the reminder of the noise resulting in an improvement in PC1. The second and third PC's were cleaned after applying the median filter a second time, however the still substantial amount of noise made both components unusable. The same procedure of noise determination we repeated on the colour Philip Island data, again showing in high NSR of 28% and using principal component analysis revealed large noise factor in components 2 and 3 . The 1st component , again as expected, produced the 'cleanest' image. All data sets showed similar NSR's and it was therefore decided that the data was of unacceptable quality and no further processing was undertaken. As the 2nd principal component shows (see figure 2) the noise throughout the image has a regular pattern. As a result it may regular pattern. As a result it may be that a Fast Fourier Transform (FFT) algorithm could be applied to the data to try and reduce this noise. Figure 2: 2nd principal component Flinders Ranges. Electronic Scanned Imagery: The electronically scanned image produced an image with a NSR of 5% the lower NSR made the imagery derived from this scanning technique more suitable for image processing techniques. The spatial resolution of the scanned images is governed by the resolution of the scanner. As the photographs were acquired at different scales a direct comparison of the resolution is difficult. Contrast Enhancement: The technique used to enhance the image was 'Contrast Variance' (Harris, 1977) or 'Statistical differencing' (Niblack, 1986) which is an algorithm useful for images where there is a large regional dynamic range coupled with a large regional variation in contrast level (Harris, 1977). This enhancement can be particularly effective for grey scale imagery with high dynamic range but; little local variation (Harrison and Jupp, 1990). The technique employs a high filter a reduce the local average to zero for all regions of the image and then applies a gain factor equal to the reciprocal of the local standard deviation to produce an image in which all local regions have equal variance (Harris, 1977). This technique improved the single channel Warrenbyne 1:25,000 monochrome photography by enhancing the smaller detail within the image. On the Flinders Ranges data an improvement was also noted showing that the technique is of benefit to multispectral images. The Phillip Island 1:10,000 imagery, did not improve using the technique due to the large amount of high frequency data already in the image. Application of the imagery Flinders Ranges: The Flinders Ranges data once imported was processed to enhance two types of features, namely spectral and structural information. To Enhance spectral information the data was linearly stretched to improve the contrast of the image. Using an interband decorrelation technique known as image logarithmic residuals (Fraser et al, 1986) the data was processed to enhance subtle spectral features. This technique has been successfully applied to Thematic Mapper data sets to enhance geological Information. The enhanced imagery showed improved lithological discrimination. A Stereo pair of aerial photographs over the Flinders Ranges area were also digitised, as single panchromatic channels. Both of these images were processed with a high pass filter to enhance structural features. These were displayed in anaglyph format where one photo is displayed on the blue colour gun, and its stereo pair on the red colour gun. Using anaglyph glasses a stereo image can be perceived . This enables on screen interpreation of the imagery . One major advantage of this method is that structural features can be digitised directly into the system, while viewing the imagery in stereo, using a GIS interface. Since this structural information is stored in the GIS as vector data it can be rectified and output as maps very quickly and over layed with other data sets such as lithological boundaries. Warrenbayne: The major advantage of the digitising technique for the Warrenbayne area is the production of accurate image maps of the area. At present the only maps in the area suitable for mapping dryland salinity are mosaiced photo maps at 1:5,00 scale derived from the 1:25,000 photographs. These maps are far from satisfactory with positional errors of up to 200m. A control network was established using the Global Positioning System (GPS) and photogrammetric techniques. In this way control points with an accuracy of + 0.8 metres in X,Y and Z where obtained . This control was used to rectify both airborne scanner imagery and also the digital aerial photographs of the area. The digital aerial photograph images were then mosaiced together to create an image base for the area. The rectification of the 1: 25,00 air photographs using this GPS control allowed the production of planimetrically accurate base maps. Phillip Island The Phillip Isalnd image was rectified to a UTM grid using an affine transformation to an accuracy of less than a pixel. Resampled to 5 and 10 metres the image was used as an objective base for vector data. The advantage of this type of image is that it gives a detailed portrayal of the ground, without any cartographic generalisation (as on a map) or sensor rasterisation that occurs on remotely sensed data. As a result the image can be used as a detailed image base for the overlay of a wide variety of geographic information such as roads and other vector data. This permits the objective. Image background to be analysed in association with the overlayed vector data. An additional advantage is that the imagery, due to the lack of generalisation, can be used by a wide variety of disciplines. Town planners, GIS personnel and local councils can 'relate' to aerial photography as 'more accessible' and 'less technical' data source. Conclusion This study into the feasibility of using digitised air photographs as an additional method of data capture has generally been successful. In this project the frame grabbed imagery was unusable. However other methods of frame grabbing imagery are available and it is quite possible that satisfactory result could be achieved using a different technique. Using the electronic scanner imagery the results for the particular case studies were encouraging with useful map products being generated. This paper has shown, through the use of selected case studies, that there are certain applications where image processed digitised aerial photographs provide information not otherwise available. Acknowledgement This research was carried out at the RMIT Centre for Remote Sensing, Victoria university of Technology, Australia. The authors express their gratitude to Craig Hill, Lucy Minato and Megan Allen for their gratitude to Craig Hill, Lucy Minato and Megan Allen for their assistance in the preparation of the paper. References Bosma, M., Drummond, J., Raidt, B. (1989) A Preliminary Report on Low-cost Scanners. I.T.C. Journal No. 2, 1989 PP 115-120. Bosma, M., Drummond, J.E. (1989 Low - cost Scanner Survey for Map-taking Applications .Working party report (2nd Draft) (available from ITC Enschede). Pp 135. Ellis, G.P. Fraser, D.D. (1990). Large Scale Mapping Using image Aalysis Proc. 5th Australian Cartographic Conference, Darwin, 1990. pp 14-25. Corblet, K.P. (ed) (1989) Landsat Data Users Notes. Eosat Publications. Vol. 4, No.4 Fraser, D.D. Mackenzie, M., Bellamy, G., Ellis, G.P. (1990) Mapping Dryland Salinity Recharge Areas using EM conductivity Mesurements and Airborne Scanner Imagery. Fifth Australian Soil Conservation Conference; working papers: Information systems and Remote Sensing papers, Perth, 1990. pp 35-52. Fraser, S.J., Horsfall, C.L. Gabell, A.R. Huntington, J.F., Green, A.A. (1986) Targeting Hydrothermal Alternation Systems in North Queensland using Aircraft Thematic Mapper Data. Proc. 4th Australsian Remote Sensing Conference Adelaide, 1986. pp 340- 351. Harris, J.L Snr. (1977) Constant Variance Enhancement a Digital Processing Technique Applied Optics. Vol. 16, No. 5 pp 1268-1271. Harrison, B.A Jupp, D.L.B. (1990) Introduction to image processing CSIRO Canberra, 1990. Harrison, R. (1990) New Tool for Audio-Visual Productions AudioVision and Prosound. June, 1990. Niblack, W. (1986) An Introduction to Digital Image Processing. Prentice-Hall International New Jersey. Rollings, N.M. (1985) Micaceous Hematite Mineralisation and Geology of the Warrakimbo Gorge, South west Flinders Ranges , South Australia. Honours Thesis. University of Adelaide. Rollings, N.M. (1988) Image processing of Landsat Thematic Mapper Data for Geological Investigations on the Western Flinders Ranges of South Australia, with Specific Application to the Mapping of Carbonates and Micaceous Hematite Exploration. Masters Thesis. University of New South Wales. Warner, W.S. (1989) A complete Small-Format Aerial Photography System for GIS Data Entry. ITC Journal No. 2 1989 pp 121-129.