the author reviews a number of parallel-processing projects which use processing power, memory and input/output of image data for image analysis. the systems covered include ILLIAC, CLIP, MPP, the cytocomputer, SNAP, ...
详细信息
the author reviews a number of parallel-processing projects which use processing power, memory and input/output of image data for image analysis. the systems covered include ILLIAC, CLIP, MPP, the cytocomputer, SNAP, and PAPIA and II. Issues in parallelism reviewed include types of parallelism, types of operation, complexity of processing elements, and the different paradigms of human vision and their application and implementation.< >
the speed-up obtained by the use of multiprocessor systems is of major importance for numerical applications involving the solution of large dense systems of linear equations. We are interested here in the performance...
详细信息
An efficient skeletonizing algorithm is presented for the hexagonal grid. the skeleton has unit width, except at crossings and in regions of the shape having even width. Otherwise the skeleton has all the properties g...
详细信息
ISBN:
(纸本)0818608781
An efficient skeletonizing algorithm is presented for the hexagonal grid. the skeleton has unit width, except at crossings and in regions of the shape having even width. Otherwise the skeleton has all the properties generally required for correct skeletons. It includes all local maxima, so complete recovery of the original shape is obtained by using the reverse distance transformation. the algorithm uses only local operations. thus it can be performed both on sequential and parallel computers. In the sequential case, described here, only three passes through the image are necessary, two to compute the distance transform and one for the identification of skeletal pixels.
the definition of a novel kind of special-purpose integrated architecture which is derived from a functional decomposition of low level image processingalgorithms is discussed. algorithms are described based on funct...
详细信息
the definition of a novel kind of special-purpose integrated architecture which is derived from a functional decomposition of low level image processingalgorithms is discussed. algorithms are described based on functional programming proposed by J. Backus (1978), so that programs are expressions to be transformed in dataflow graphs. Due to image processing algorithm properties, most such data flow graphs are regular and can be implemented on a mesh-connected programmable dataflow functional computer.< >
A method is presented for the segmentation and description of 3-D objects in a range image. the image is segmented in two parallel ways: by region classification based on surface normal analysis and by edge detection ...
详细信息
ISBN:
(纸本)0818608781
A method is presented for the segmentation and description of 3-D objects in a range image. the image is segmented in two parallel ways: by region classification based on surface normal analysis and by edge detection using a relaxation process. the two results are combined by a small rule-based system to obtain a more reliable segmentation. Features of the 3-D object at different levels are extracted and a surface attributes graph is constructed. this kind of view-independent description of 3-D objects is effective and robust for object recognition and location. the objects considered are industrial parts composed of planes and quadric surfaces. the range image is provided by a laser scanner.
A parallelthinning algorithm of C.M. Holt et al. (1987) is compared with an algorithm of D. Rutovitz (1966) and one by T.Y. Zhang and C.Y. Suen (1984). Analyses and experiments show that the Holt algorithm is similar...
详细信息
A parallelthinning algorithm of C.M. Holt et al. (1987) is compared with an algorithm of D. Rutovitz (1966) and one by T.Y. Zhang and C.Y. Suen (1984). Analyses and experiments show that the Holt algorithm is similar to the Rutovitz algorithm. A heuristic modification to Rutovitz' algorithm is also proposed and the modified algorithm is faster than Holt's algorithm.
Preprocessing, which plays an important role in pattern recognition, is investigated from the point of view of real-time processing, and a set of one-pass processingalgorithms are described. All of the algorithms, wh...
详细信息
ISBN:
(纸本)0818608781
Preprocessing, which plays an important role in pattern recognition, is investigated from the point of view of real-time processing, and a set of one-pass processingalgorithms are described. All of the algorithms, which are based on easy hardware implementation, only require one single cycle of parallel operations per image frame, and the operations are confined in a window of size 3*3. Computer simulation shows that the algorithms are efficient and easily realized in hardware.< >
the magnitude-distance and maximum-value-distance filters are analyzed, and fast realizations are proposed to implement a DRF cascade for edge detection. It is shown that the maximum-value-distance filter is well adap...
详细信息
ISBN:
(纸本)0818608781
the magnitude-distance and maximum-value-distance filters are analyzed, and fast realizations are proposed to implement a DRF cascade for edge detection. It is shown that the maximum-value-distance filter is well adapted to parallelprocessing and works two times as fast as the magnitude-distance filter. It is shown also that the adaptive gradient is an optimal estimate for edge-pixel verification after zero-crossing detection. the efficiency of the proposed algorithms is confirmed by experimental results.< >
the split-level relaxation technique is analyzed in a parallelprocessing framework. It is shown that there is much parallelism inherent in the algorithm that can be exploited. this has been confirmed by implementatio...
详细信息
the split-level relaxation technique is analyzed in a parallelprocessing framework. It is shown that there is much parallelism inherent in the algorithm that can be exploited. this has been confirmed by implementation of the algorithm on an actual multiprocessor. Although the results are good, the implementation can be made more efficient. the use of multiple queues instead of a centralized queue can reduce memory contention, particularly in large multiprocessors. An asynchronous implementation can also improve the performance.< >
Problems posed by the application of the Hough transform to a small-scale image are discussed. It is shown that the discretization of image space makes the distribution of parameters in Hough space nonhomogeneous and ...
详细信息
Problems posed by the application of the Hough transform to a small-scale image are discussed. It is shown that the discretization of image space makes the distribution of parameters in Hough space nonhomogeneous and nonequiprobable. A method of correction is presented in the case of the detection of parallel straight lines on a square scan window.< >
暂无评论