A methodology for optical flow analysis based on cepstral filtering is introduced. The power cepstrum is extended to multiframe analysis. A correlative cepstral technique, cepsCorr, is developed. It significantly incr...
详细信息
A methodology for optical flow analysis based on cepstral filtering is introduced. The power cepstrum is extended to multiframe analysis. A correlative cepstral technique, cepsCorr, is developed. It significantly increases the signal-to-noise ratio, reduces ambiguities, and it provides a predictive or multievidence approach to visual motion analysis.< >
We present a method to smooth a signal-whether it is an intensity image, a range image or a planar curve-while preserving discontinuities. This is achieved by repeatedly convolving the signal with a very small averagi...
详细信息
We present a method to smooth a signal-whether it is an intensity image, a range image or a planar curve-while preserving discontinuities. This is achieved by repeatedly convolving the signal with a very small averaging mask weighted by a measure of the signal continuity at each point. The method is extremely attractive since edge detection can be performed after a few iterations, and features extracted from the smoothed signal are correctly localized. Hence no tracking is needed, as in Gaussian scale-space. This last property allows us to derive a new scale-space representation of a signal using the adaptive smoothing parameter k as the scale dimension. We then show how this process relates to anisotropic diffusion. When a large amount of smoothing is desired, we propose a multigrid implementation which reduces the computational time significantly. Given the local nature of the algorithm, we also propose a parallel implementation: the running time on a 16K Connection Machine is three orders of magnitude faster than on a serial machine. We then present several applications of adaptive smoothing: edge detection, range image feature extraction, corner detection, and stereo matching. Examples are given throughout the text using real images.
New piecewise-linear basis functions called PLh functions are proposed. Using the PLh functions a new digital signalprocessing transform, PLh transform, is derived. The PLh transform has an attractive feature of the ...
详细信息
New piecewise-linear basis functions called PLh functions are proposed. Using the PLh functions a new digital signalprocessing transform, PLh transform, is derived. The PLh transform has an attractive feature of the economy of computational effort both in computation time and in storage size. Thus the new transform is suitable for imageprocessing and other applications in which a large number of data need to be processed. Application of the PLh transform to edge detection of two dimensional image is also discussed.
A dynamic grey-level thresholding scheme for the binary image segmentation is proposed. By the use of edge information and region-growing methods the effects of uneven illumination, noise and blurring are diminished, ...
详细信息
A new edge-adaptive low-pass filter is proposed. This allows substantial removal of noise while maintaining clarity of edges and areas of high detail. Some results are presented and compared with those obtained by a s...
详细信息
New piecewise-linear basis functions called PLh functions are proposed. Using the PLh functions a new digital signalprocessing transform. The PLh transform, is derived. The PLh transform has an attractive feature of ...
详细信息
New piecewise-linear basis functions called PLh functions are proposed. Using the PLh functions a new digital signalprocessing transform. The PLh transform, is derived. The PLh transform has an attractive feature of economy of computational effort both in computation time and in storage size. Thus the new transform is suitable for imageprocessing and other applications in which a large number of data need to be processed. Application of the PLh transform to edge detection of a two-dimensional image is also discussed.< >
The authors introduce two multiscale optical imageprocessing systems, a centroid scale-space processor and an optical wavelet processor. A.P. Witkin (1984) introduced the idea of scale-space, where information is pre...
详细信息
The authors introduce two multiscale optical imageprocessing systems, a centroid scale-space processor and an optical wavelet processor. A.P. Witkin (1984) introduced the idea of scale-space, where information is presented on a coordinate system with continuous spatial and scale axes. His representation consists of a convolving the input signal with a Laplacian of a Gaussian point spread function and retaining the zero-crossings (which correspond to edges) as scale (width of the Gaussian) is varied. These edge maps form continuous contours in scale-space. The optical implementation of the system is discussed. The centroid scale-space map (CSSM) is constructed. This technique is extended to 2D input functions by using correlation techniques. The optical system which computes a 2D wavelet transform is shown. One special advantage is that the system complexity is independent of the scale parameter. This gives the flexibility to tailor the scale parameter to the particular application.< >
In order to manipulate SHD (super high definition) images, a novel parallel processing unit called HiPIPE (Highly Parallel imageprocessing Engine) is developed as a project of NOVI-II. The engine is connected to an S...
详细信息
In order to manipulate SHD (super high definition) images, a novel parallel processing unit called HiPIPE (Highly Parallel imageprocessing Engine) is developed as a project of NOVI-II. The engine is connected to an SHD image display unit and an image data storage unit. Extremely high computational power is obtained by a multicomputer type parallel processing technique. 128 processing elements are connected by a mesh network. Various image coding schemes are carefully explored from the viewpoint of parallel processing, and the problem of processor connections is examined. A novel load-balancing technique, called 2-dimensional butterfly data shuffling, is developed and implemented. The current version of HiPIPE uses just 128 scalar processing elements and has more than twice the power of a single-processor CRAY-2 for a still SHD image coding task.< >
暂无评论