Enhancement of lmage Resolution in Digital Photogrammetry

In recent years, considerable developments have occurred in the field of digital photogrammetry. These have been due mainly to increases in computing power, the refinement of featureand area-based image matching algorithms and the reduction in the cost of equipment capable of producing near real-time images in digital format. A major limitation to the widespread application of digital photogrammetry concerns the small format size of the cm sensor itself and, consequently, the number of pixels on the sensor being limited in number. Much time and effort has been expended trying to improve coverage through hardware solutions such as producing imaging sensors with increased numbers of pixels. An alternative software solution is offered in this paper. An algorithm which combines several digital images, the photogrammetric technique of area-based image matching, and a rigorous mathematical solution to increase the effective number of pixels is described. The resolution of the final composite image is enhanced relative to its constituent images. Introduction Photogrammetry is a discipline which has derived benefit from the developments in digital technology. Methods and techniques are continually being devised to take advantage of emerging technologies, and new and diverse applications are being undertaken which were not feasible with traditional photogrammetric techniques. The advancement of digital image technology and digital photogrammetry has been ofbenefit to applications in a wide range of fields, including industrial measurement, archaeological, architectural, astronomical, medical, GIS updating, closerange and aerial mapping, and forensic applications. Industrial measurement applications are diverse, ranging from the measurement of the surfaces of aircraft wings and wind turbine blades (Robson and Setan, 1996), the characteristics of train body surfaces (Kochi et al., 1996) to reverse engineering from physical models (Petran and Krzystek, 1996). Medical applications use images not only from CCD cameras, but also from xrays and other medical imaging sensors. Examples of such applications include the measurement of the changes to soft tissue after facial surgery (Gabel and Kakoschke, 1996) and the verification of the position of teeth during and after orthodontic treatments (Hdflinger, 1996). Virtual reality environments are being modeled from real scenes using digital photogrammetry and, coupled with data from laser scanning, are being used for industrial, medical, and training applications (El-Hakim et al., 1996). Forensic imaging Department of Civil, Surveying and Environmental Engineering, University of Newcastle, Callaghan, NSW 2308, Australia ([email protected]). K. McIntosh is presently with the Department of Engineering, Faculty of Civil Engineering, Technion-Israel Institute of Technology, Haifa, 3200, Israel ([email protected]). may involve the analysis of existing imagery such as old photographs which are digitized and then processed using image enhancement and recognition techniques to investigate crimes (Robertson, 1998). New applications of digital photogrammetry are being reported at an increasing rate. Many applications, particularly close-range, require the speed and on-line capabilities of analog CCD cameras, or the portability and flexibility of digital still cameras. The main objective of digital photogrammetry remains the same as that of traditional film-based techniques: to obtain accurate spatial information about remotely sensed objects. Unfortunately, digital cameras with high numbers of pixels are expensive and inaccessible to many users. Often, a cheaper camera with a smaller format, and less pixels, will be used. If the same object area is captured, the resulting image may be at a lower than desirable resolution, compromising the eventual accuracy for the re-creation of the object. Areas where digital photogrammetry can be efficiently used are often limited by the resolution of the imagery. The resolution can also affect the visual quality of the results and the precision of classifications made from the imagery. This limitation has been noted by several researchers (for examples, see Uffenkamp (1993), Wong and Obaidat (1994) and Motala (1997)). Revlew of lmage Enhancement Literature The objective of image enhancement is to produce an image which is more suitable for an application than the original image (Gonzalez and Wintz, 1987), thus improving the original image to give better visualization (Weeks, 1996) or increased accuracy in classifications or measurements. Hardware and software solutions are the two ways this enhancement can be achieved. Hardware solutions to produce enhanced resolution include increasing the size of the sensor, increasing the number of pixels (photosites) on the sensor, decreasing the pixel size, and modifying a camera to move the sensor by known amounts. Software solutions include using one or more low resolution images to interpolate or solve for a higher resolution image when there are initially unknown amounts of shift between images. Hardware Solutions There have been many investigations into mediumand highresolution cameras which are commercially available and, as previously mentioned, are relatively expensive (Bbsemann et al., 1990; Luhmann, 1990; Maas and Niederost, 1997). Peipe (1995) presented an investigation of the Kodak DCS460, which is a digital still camera with a 3000 by 2000-pixel sensor. The Photogrammetric Engineering & Remote Sensing Vol. 67, No. 6, June 2001, pp. 741-749. 0099-1112/01/6706-741$3.00/0