<title>Direct Electronic Fourier Transform (DEFT) Spectra For Terrain Feature Classification</title>

Abstract

Abstract This paper presents the experimental results (i.e., spectra) obtained with a Direct Elec­ tronic Fourier Transform (DEFT) device. Emphasis is placed on the use of DEFT spectra for terrain feature classification. These spectra indicate the potential of DEFT technology for distinguishing between natural features and cultural or man-made features. These same spec­ tra appear to have potential for distinguishing between certain subclasses of terrain fea­ tures, e.g., open fields, bodies of water, and woods. A potential application of DEFT spectra could be the preliminary analysis of aerial imagery to automatically flag certain photographs for subsquent detailed analysis by a human photointerpreter (PI), and to auto­ matically select or reject specific photographs before digitization for mapping or other purposes. New devices use surface acoustic waves (SAW) to generate a two-dimensional limited bandwidth Fourier transform of an image in real time without the aid of a computer. These devices permit spectra analysis for a two-dimensional image as a communications engi­ neer would analyze the RF spectrum of a radio signal. Appropriate references and a brief description of DEFT technology will be presented for those unfamiliar with this advance in solid state, acousto-optic technology.IntroductionSurface acoustic wave (SAW) technology has made many advances in recent years. One such advance is the use of a SAW to interact with a conventional image, I(x,y), focused on a film of CdS to produce a Fourier transform of the image. Devices produced with this technology are referred to as Direct Electronic Fourier Transform (DEFT) devices.DEFT devices were invented by Professors P. G. Kornreich and S. T. Kowel at Syracuse Uni­ versity. These devices have been brought to their present level of development through Army sponsorship based on selection by the Advanced Concepts Teams (ACT), Headquarters, Depart­ ment of the Army.A schematic of an early DEFT device is illustrated in Figure 1. A schematic of a recent DEFT device is illustrated in Figure 2.It Is recognized that there is considerable Interest in the device itself and the physics of its operation. Complete details are beyond the scope of this paper but can be found in references 1 through 12. However, a very concise description of DEFT technology Is that a propagating SAW effectively modulates the photoconductivity of a film of CdS. This modula­ tion Is spatial and temporal. If an image, I(x,y), Is focused on the CdS, an alternating current (ac) as well the usual direct current (dc) is generated and detected. If the wavelength of the SAW corresponds to the spatial wavelength in the image, then the detected ac current has a frequency corresponding to the wavelength of the SAW. The usual relation­ ship, f = v/X , applies. All variables refer to the SAW. This ac current is proportional to the product of the intensity distribution of the image and the SAW, I.e.,// I(x,y) exp -2irj{ (x/Xx) + (y/X )}dx dy. If the frequency of the SAW is swept, then there Is a series of ac currents corresponding to the range of frequencies swept and with ampli­ tudes determined by I(x,y). This series of ac currents is proportional to the Fourier transform of the image. This ac signal can be displayed on a network analyzer, oscillo^ scope, or plotter. DEFT devices can be considered to be active filters which pass ac fre­ quencies corresponding to the spatial frequencies of an Image. As with any electronic de­ vice, they have a specific bandwidth. They also scan, sample, and transform the Image in a specific manner. Hence, it is appropriate to refer to their output as the DEFT spectrum of the Image.A distinguishing feature of the DEFT method is that the Fourier transform is performed in a random access mode. That is, the whole image is viewed and line after line of transform data is taken. The result is the conversion of a two-dimensional space domain input to a two-dimensional frequency domain output. Standard digital methods for calculating the fast Fourier transform (FFT) of an image generate a two-dimensional transform with Fourier fre­ quencies in all directions. However these frequencies cannot be randomly accessed until the whole transform has been calculated.