Moga, and M. Gabbouj, Parallel Image Component Labeling with Watershed Transformation. IEEE Transaction on Pattern Analysis and Machine Intelligence, May 1997, Vol 19, No 5.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....Simulation experiments have been conducted to verify the load balancing algorithm and measure its performance which is summarized in the following section. 5.2.Simulation Experiments We built a discrete time event simulation model of HNOW to test the proposed SDLB algorithm. We used OMNET [19] which provides the simulation environment and programmed the model using visual C . First to verify our simulation model we made exhaustive tests with deterministic values and compared the results to the expected value. Figure 4 shows a case in which 5 identical workstations are assigned random ....

N.A. Moga and M. Gabbouj, "Parallel Image Component Labeling with Watershed Transformation," IEEE Trans. on Pattern Analysis and Machine Intelligence, Vol. 19 No. 5, pp. 441--450, 1997.

....ordered queues. 5.2.3. Hill climbing by ordered queues combined with a connected component operator Parallelization of the hill climbing algorithm combined with a connected component operator has been considered by Bieniek et al. 6] using the local condition of De nition 3. 7, and by Moga et al. [27, 30]. We rst describe the former approach [6] The main idea is to solve the watershed problem independently on all subdomains without synchronization. Instead temporary labels are assigned to pixels which will be ooded from adjacent subdomains. The boundary connectivity information is stored in a ....

....according to De nition 3.7. a) original image; b) a watershed segmentation of the complete image; c) result after step 2 of the parallel algorithm with two subdomains. d) result after step 3 of the parallel algorithm with two subdomains. A similar approach was used by Moga et al. [27,30]. The di erence with the approach in [6] is that for non minima plateaus which are shared by several processors the globally correct lower distance values are computed before ooding, instead of during ooding, so that no relabelling of wrongly labelled higher neighbourhoods of plateaus is ....

[Article contains additional citation context not shown here]

Moga, A. N., and Gabbouj, M. Parallel image component labeling with watershed transformation. IEEE Trans. Patt. Anal. Mach. Intell. 19, 5 (May 1997), 441-450.

....impulsive noise is present [1] 2] Otherwise, both energies give quite similar results. And, since the activity image is recalculated as the scale increases, it also becomes more cleaned. To obtain the final segmentation the (cleaned) activity image is then fed through a watershed algorithm [5][6]. In this paper we use a floating point implementation that consists of two steps. First, all pixels that are below a certain threshold level, i.e. the flooding or drowning level, are merged into attraction basins [5] and second, the remaining pixels are merged with their neighbours in the 1 at ....

N.A. Moga and M. Gabbouj, Parallel Image Component Labeling with Watershed Transformation, IEEE Trans. on Pattern Analysis and Machine Intelligence, Vol. 19 No. 5, pp. 441--450, 1997. 2 INRIA--Syntim c fl; see http://www-syntim.inria.fr/syntim/analyse/images-eng.html

....pixels. As the diffusion proceeds, the noise in the image is gradually reduced while the edges are kept as sharp as possible. In the watershed segmentation step the activity image is used as a topographic surface whose watersheds, i.e. the regions bordered by the mountain rims, form the segments [5, 6]. Traditional watershed segmentation algorithms use a quantized activity image that, to reduce the computation time, must not contain too many levels. Our implementation [1, 2] takes an activity image with a floating point activity value per pixel as input, circumventing the disadvantage of ....

N.A. Moga and M. Gabbouj, Parallel Image Component Labeling with Watershed Transformation, IEEE Trans. on Pattern Analysis and Machine Intelligence, Vol. 19 No. 5, pp. 441--450, 1997.

.... spanning forest (MSF) of the graph is computed, constraining that every tree in the MSF contains exactly one marked vertex (region) see [12] For the first part of the problem, namely, parallelization of the watershed algorithm, an efficient connected components algorithm has been presented in [18]. In this paper, the algorithm is modified to additionally perform in parallel marker based region merging by means of a constrained MSF operator. The algorithm represents an extension of the classical MSF problem, namely, to find the MSF of a graph with vertices marked, in part, such that in ....

....techniques have been used in [13, 14, 15, 16] to bear a scalable, efficient parallel algorithm. Yet, another strategy of labeling connected components in parallel like in [1] was proven applicable for the watershed transformation which does not build watershed lines (0 width watershed lines) [18]. The main idea in [18] is to derive a local precedence relation between a candidate pixel and its already flooded neighbor, at which the candidate pixel will be appended. Once this relation is established, connected components are first locally correctly labeled. However, components which extend ....

[Article contains additional citation context not shown here]

A.N. Moga and M. Gabbouj, "Parallel image component labeling with watershed transformation, " IEEE Trans. Patt. Anal. Mach. Intell., vol. 19, no. 5, pp. 441--450, May 1997.

.... successfully incorporated into image analysis systems in various domains, e.g. in industry and biomedicine (segmentation of electrophoresis gels, a moving heart, 3D holographic images, road traffic analysis) 19] Starting with a successful parallel design solution of the watershed algorithm [6,22], further tests on different parallel machines have been performed to evaluate its portability and performance. In our efforts to find a new parallel algorithm for the watershed problem displaying concurrency, locality, modularity, data independence, and resilience to increasing number of ....

....inserted in the FIFO queue allocated to its grey level. However, because the classical algorithm implements a global and highly data dependent operation, its parallelisation is not straightforward. An efficient parallel implementation of the watershed algorithm has been extensively presented in [22], with results collected from a Cray T3D parallel computer. In order to prove the portability of the design solution on different massively parallel computers and hence, of its performance, the algorithm has been tested on several parallel machines. Before presenting the results, a short ....

[Article contains additional citation context not shown here]

A.N. Moga and M. Gabbouj. Parallel image component labeling with watershed transformation. IEEE Transactions on Pattern Analysis and Machine Intelligence, 19(5):441--450, May 1997.

No context found.

Moga, and M. Gabbouj, Parallel Image Component Labeling with Watershed Transformation. IEEE Transaction on Pattern Analysis and Machine Intelligence, May 1997, Vol 19, No 5.

No context found.

A.N. Moga and M. Gabbouj, Parallel image component labeling with watershed transformation, IEEE Transactions on Pattern Analysis and Machine Intelligence, 19(5):441-450, May 1997 http://www.informatik.uni-freiburg.de/papers/lmb/mo_pami97.ps.gz.

No context found.

A. N. Moga and M. Gabbouj. Parallel image component labeling with watershed transformations. IEEE Transactions on Pattern Analysis and Machine Intelligence, 19(5):441--450, 1997.

No context found.

Moga, A. N., and Gabbouj, M. Parallel image component labeling with watershed transformation. IEEE Trans. Patt. Anal. Mach. Intell. 19, 5 (May 1997), 441--450.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC