Search

We report a wavelength-compensated three-channel (RGB)joint transform correlator (JTC) for color pattern recognition using a ferroelectric liquid-crystal spatial light modulator (SLM) operating in binary pure phase modulation. We apply a previously reported time-multiplexing technique useful in creating wavelength-compensated diffraction patterns, based on the synchronization of properly scaled diffraction masks with the input wavelength selection obtained by applying a rotating RGB color-filter wheel to an Ar--Kr laser. The application of this technique to a JTC architecture permits real-time color object detection. In order to achieve a high light efficiency for the correlation process, we combine the design of zero-order joint power spectra in all color channels with the selection of a certain polarization configuration of the SLM, producing a broadband phase-only modulation. Excellent experimental results demonstrating color-object detection are provided. OCIS codes: 070.5010, 100.5010, 070.4550, 100.4550, 070.6120.

Taking advantage of small aberrations, high efficiency, and compactness, we developed a new, to our knowledge, design procedure for a binary zone-plate array (BZPA) and applied it to a parallel joint transform correlator for the recognition of the human face. Pairs of reference and unknown images of faces are displayed on a liquid-crystal spatial light modulator (SLM), Fourier transformed by the BZPA, intensity recorded on an optically addressable SLM, and inversely Fourier transformed to obtain correlation signals. Consideration of the bandwidth allows the relations among the channel number, the numerical aperture of the zone plates, and the pattern size to be determined. Experimentally a five-channel parallel correlator was implemented and tested successfully with a 100-person database. The design and the fabrication of a 20-channel BZPA for phonetic character recognition are also included.

A computer-vision system is proposed that was designed based on computational reflection and control-system theory. The system's framework has enough semantic information to analyze its own behavior. Hence, it is capable of monitoring its own performance and adapt accordingly to varying circumstances and requirements. The computer-vision system may be useful for aerial surveillance and can be applied to medical imaging and intelligent transportation systems.

The problem of finding the closest point in high-dimensional spaces is common in pattern recognition. Unfortunately, the complexity of most existing search algorithms, such as k-d tree and R-tree, grows exponentially with dimension, making them impractical for dimensionality above 15. In nearly all applications, the closest point is of interest only if it lies within a user-specified distance[Epsilon]. We present a simple and practical algorithm to efficiently search for the nearest neighbor within Euclidean distance [Epsilon]. The use of projection search combined with a novel data structure dramatically improves performance in high dimensions. A complexity analysis is presented which helps to automatically determine [Epsilon] in structured problems. A comprehensive set of benchmarks clearly shows the superiority of the proposed algorithm for a variety of structured and unstructured search problems. Object recognition is demonstrated as an example application. The simplicity of the algorithm makes it possible to construct an inexpensive hardware search engine which can be 100 times faster than its software equivalent, A C++ implementation of our algorithm is available upon request to search@cs. columbia.edu/CAVE/. Index Terms - Pattern classification, nearest neighbor, searching by slicing, benchmarks, object recognition, visual correspondence, hardware architecture.