An image processor implementing algorithms using characteristics of phase spectrum of two-dimensional Fourier transformation

A series of new image processing algorithms are emerging which specifically focuses on the characteristics of the phase components obtained from two-dimensional Fourier transformation. Phase only correlation (POC) is the most fundamental one, and its features include superior discrimination capability over the ordinary correlation, robustness, and one hundredth grade sub-pixel resolution. Applications of the algorithms range from measurements of displacement, rotation, and distance of a target, and detection of scratches or stains on an image. In this paper, we discuss an image processor that executes the algorithms at the speed fast enough for real-time image processing applications. The processor executes 256/spl times/256 pixels POC in 27.15 ms.