A High Performance 3d Exact Euclidean Distance Transform Algorithm for Distributed Computing

Torelli, Julio Cesar; Fabbri, Ricardo; Travieso, Gonzalo; Bruno, Odemir Martinez

doi:10.1142/s0218001410008202

Cited by 10 publications

(6 citation statements)

References 24 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…A Distance Transform (DT) measures the distance in pixels from a pixel to the edge of the region of interest regardless the considered direction [57][58][59][60][61][62][63]. The DT allows finding useful geometric information within images; having this information, regions of interest can be found for many applications, including medicine etc.…”

Section: Fast Distance Transform (Fdt)mentioning

confidence: 99%

“…[57,60]. A widely used metric for the DT is the Euclidean distance for the extraction of geometric information [59,61,62], in fact, the algorithm of the Euclidean distance transform has been parallelized so that it may be processed even faster [59]. As one can see, TD is a very useful tool in image processing for data extraction.…”

Section: Fast Distance Transform (Fdt)mentioning

confidence: 99%

See 1 more Smart Citation

New Model of Heteroasociative Min Memory Robust to Acquisition Noise

et al. 2022

View full text Add to dashboard Cite

Associative memories in min and max algebra are of great interest for pattern recognition. One property of these is that they are one-shot, that is, in an attempt they converge to the solution without having to iterate. These memories have proven to be very efficient, but they manifest some weakness with mixed noise. If an appropriate kernel is not used, that is, a subset of the pattern to be recalled that is not affected by noise, memories fail noticeably. A possible problem for building kernels with sufficient conditions, using binary and gray-scale images, is not knowing how the noise is registered in these images. A solution to this problem is presented by analyzing the behavior of the acquisition noise. What is new about this analysis is that, noise can be mapped to a distance obtained by a distance transform. Furthermore, this analysis provides the basis for a new model of min heteroassociative memory that is robust to the acquisition/mixed noise. The proposed model is novel because min associative memories are typically inoperative to mixed noise. The new model of heteroassocitative memory obtains very interesting results with this type of noise.

show abstract

Section: Fast Distance Transform (Fdt)mentioning

confidence: 99%

Section: Fast Distance Transform (Fdt)mentioning

confidence: 99%

New Model of Heteroasociative Min Memory Robust to Acquisition Noise

et al. 2022

View full text Add to dashboard Cite

show abstract

“…Saito and Toriwaki [19] proposed a sequential algorithm based on dimensionality reduction, since the squared Euclidean distance value is separable, the distance transform can be computed along each principal direction. Recently, Torelli et al [20] implemented Saito's algorithm on a cluster using MPI. However, Saito's algorithm has a time complexity of O(n 3 ) and it will result in a poor performance in practice.…”

Section: A Exact Edtmentioning

confidence: 99%

A Work Efficient Parallel Algorithm for Exact Euclidean Distance Transform

Manduhu

Jones

2019

IEEE Trans. on Image Process.

View full text Add to dashboard Cite

A fully-parallelized work-time optimal algorithm is presented for computing the exact Euclidean Distance Transform (EDT) of a 2D binary image with the size of n × n. Unlike existing PRAM and other algorithms, this algorithm is suitable for implementation on modern SIMD architectures such as GPUs. As a fundamental operation of 2D EDT, 1D EDT is efficiently parallelized first. Specifically, the GPU algorithm for the 1D EDT, which uses CUDA binary functions such as ballot(), ffs(), clz() and shfl(), runs in O(log32n) time and performs O(n) work. Using the 1D EDT as a fundamental operation, the fullyparallelized work-time optimal 2D EDT algorithm is designed. This algorithm consists of three steps. Step 1 of the algorithm runs in O(log32n) time and performs O(N) (N =n 2) of total work on GPU. Step 2 performs O(N) of total work and has an expected time complexity of O(logn) on GPU. Step 3 runs in O(log32n) time and performs O(N) of total work on GPU. As far as we know, this algorithm is the first fully-parallelized and realized work-time optimal algorithm for GPUs. Experimental results show that this algorithm outperforms prior state-of-theart GPU algorithms. Index Terms-work-time optimal parallel algorithm, SIMD architecture, 1D EDT, 2D EDT, binary operations on GPU.

show abstract

“…Various scientific areas such as mathematics, biology, medicine, computing [2], [3] have a need for HPC or HTC infrastructure. High performance of computing tasks is achieved by parallel processing and using distributed memory for communication.…”

Section: Cluster Computingmentioning

confidence: 99%

Data-Intensive Computing Paradigms for Big Data

Lončar¹

2018

Proceedings of the 29th International DAAAM Symposium 2018

View full text Add to dashboard Cite

The development of ICT has led to enormous growth and data accumulation, and thus created the need for proper storage and processing of large data, known as Big Data. The number of data sources like mobile telephones and applications, social networks, digital television, data of various Internet objects and sensors has increased due to the development of technology and the emergence of the IoT, the evolutionary step in the development of the Internet. Analysis and proper interpretations that take place on the latest distributed platforms are key to data-intensive systems from which feedback can be gained in areas such as industry, finance, healthcare, science and education. Distributed computing paradigms are a fundamental component of research and innovation for e-infrastructures with the intent of providing advanced computing, storage resources and network connectivity necessary for modern and multidisciplinary science and society. The aim of this paper is to provide a systematic overview and address challenges of distributed computing paradigms that marked and brought revolution into computing science and are able to store and process large amounts of data. As an example of a big data source, the world's largest particle detector, CERN's LHC is analyzed.

show abstract

A High Performance 3d Exact Euclidean Distance Transform Algorithm for Distributed Computing

Cited by 10 publications

References 24 publications

New Model of Heteroasociative Min Memory Robust to Acquisition Noise

New Model of Heteroasociative Min Memory Robust to Acquisition Noise

A Work Efficient Parallel Algorithm for Exact Euclidean Distance Transform

Data-Intensive Computing Paradigms for Big Data

Contact Info

Product

Resources

About