Parallel Thinning Research Articles

Implementation of parallel thinning algorithms using recurrent neural networks

The use of recurrent neural networks for skeletonization and thinning of binary images is investigated. The networks are trained to learn a deletion rule and they iteratively delete object pixels until only the skeleton remains. Recurrent neural network architectures that implement a variety of thinning algorithms, such as the Rosenfeld-Kak algorithm and the Wang-Zhang (WZ) algorithm, are presented. A modified WZ algorithm which produces skeletons that are intuitively more pleasing is introduced.

Skeleton-Based Design Analysis of Near Net-Shape Products

Summary In this paper, we present a parallel thinning algorithm for design analysis of near net shape products. The parallel thinning algorithm allows extraction of global topology of a three dimensional geometry. In this scheme, the geometry is digitized and a skeleton line of the geometry is digitized by preserving the Euler characteristics and connectivity. The extracted skeleton allows a design evaluation system to utilize heuristic manufacturing knowledge base. Finally, we construct a skeleton model and present its application to casting and forging design analysis.

A new one-pass parallel thinning algorithm for binary images

Abstract A new parallel thinning algorithm for binary images is proposed. This algorithm employs template matching to remove the edge points of an object shape in a binary image iteratively. Difficulty of terminating the thinning process without excessive erosion is solved without dividing an iteration into several passes. With a newly designed set of matching templates, the proposed algorithm obtains skeletons with the properties of perfectly 8-connected, noise-insensitive, and topologically equivalent to the original object shape without excessive erosion.

A neural network hardware model for image thinning

A new parallel thinning algorithm was proposed. This paper suggests a set of 5x5 cloning templates of cellular neural network for image thinning. The principle of the templates design and hardware model are proposed in the paper. Althouth the connected weights between the neurons are asymmetric, it is shown that the network is stable, and it can be easily realized.

Further considerations of a 1‐subcycle parallel thinning algorithm

In this paper, we present two improved versions of a pseudo 1-subcycle parallel thinning algorithm which we proposed earlier [1] and describe a twostage structure to realize the 1-subcycle parallel algorithm. The purpose of the first improved algorithm is to produce the P8ET (perfect 8-curve excluding T-junction)-type thin line which is well-defined in general. This algorithm involves local branches containing T-junctions or quasi T-junctions. The second improved algorithm obtains an isotropic skeleton of an L-shape pattern, such that the visual quality of a skeleton is more satisfactory. In this algorithm, another set of thinning conditions especially designed is also involved. We also describe a two-stage structure which consists of a thinning table and a control unit. The thinning table is used to provide the attributes of an input 3×3 local pattern. The control unit is used to check the removal of the center 1 pixel of this local pattern. The inputs of the control unit also comprise the outputs of other neighboring thinning tables. This structure can exactly implement the original 1-subcycle parallel algorithm. The above-mentioned algorithms have been implemented on this realized structure in this paper. Several experiments confirm that the improved algorithms can produce the desired effective thin line, and also show that the realized structure is feasible and practicable.

Comparative evaluation of fast thinning algorithms on a multiprocessor architecture

Abstract A comparative evaluation of some fast parallel thinning algorithms for elongated shapes on an overlapped shared memory multiprocessor architecture (OSMMA) is presented. The evaluation is based on the execution time and multiprocessor speed-up of the algorithms, as well as the quality of the skeletonized image structures. Issues relating to efficient algorithm implementation on the distributed memory architecture and refinements in the algorithms for better results are also discussed. The algorithms implemented include our own thinning technique based on the concept of object edges width compression that achieves efficient results while significantly decreasing execution timings. Experimental results for some real images are presented.

Fast fully parallel thinning algorithms

Three new fast fully parallel 2-D thinning algorithms using reduction operators with 11-pixel supports are presented and evaluated. These are compared to earlier fully parallel thinning algorithms in tests on artificial and natural images; the new algorithms produce either superior parallel computation time (number of parallel iterations) or thinner medial curve results with comparable parallel computation time. Further, estimates of the best possible parallel computation time are developed which are applied to the specific test sets used. The parallel computation times of the new algorithms and one earlier algorithm are shown to approach closely or surpass these estimates and are in this sense near optimally fast.

Comments on “a parallel-symmetric thinning algorithm” by Bourbakis

In a recent paper (Bourbakis, Pattern Recognition 22, 387–396 (1989)) a thinning algorithm which is characterized as a parallel-symmetric thinning algorithm (PSTA) is presented. There are several concerns regarding this algorithm which bring into question its utility as a parallel thinning algorithm. In this note it is shown that PSTA is fundamentally the parallel application of two largely sequential components and is most aptly characterized as a sequential algorithm; that it produces unacceptable geometric results for many images; and that it fails to preserve connectivity properties for a large class of images.

A flexible thinning algorithm allowing parallel, sequential, and distributed application

A parallel thinning algorithm for digital patterns is presented. This algorithm can also be used for sequential thinning without the simulation of parallel computation. The mathematical background of the algorithm bases on the notion of the Euler characteristic. The proposed algorithm is simple and particularly faster than other parallel algorithms.

A Study on the Recognition and Understanding Expert System of Mechanical Drawings. A Thinning Algorithm and a Rule Interpreter for Understanding of Mechanical Drawings.

The authors are developing an intelligent recognizing and understanding system for mechanical drawings. This system is designed to extract the meanings of drawings and translate them into CAD data. Some of the characteristic features of the system are a module specialized in the image processing of horizontal and vertical lines and an image understanding module based on interpretation rules. To achieve a higher performance, a new parallel thinning algorithm and a rule interpreter are introduced. The thinning algorithm uses protective patterns, and it is expected to preserve corners and diverging points consisting of horizontal and verticallines. The rule interpreter is newly designed by analyzing the interpretation process for mechanical drawings. The thinning algorithm gives results with less distortion than a previous sequential thinning algorithm, and it is also applicable to the image data containing oblique lines and curves. The rule interpreter enhances the ability and the flexibility of image understanding, and it improves the interpretation result.

A SINGLE-LAYER NEURAL NETWORK FOR PARALLEL THINNING

A single-layer recurrent neural network is proposed to perform thinning of binary images. This network iteratively removes the contour points of an object shape by template matching. The set of templates is specially designed for a one-pass parallel thinning algorithm. The proposed neural network produce the same results as the algorithm. Neurons in the neural network performs a sigma-pi function to collect inputs. To obtain this function, the templates used in the algorithm are transformed to equivalent Boolean expressions. After the neural network converges, a perfectly 8-connected skeleton is derived. Good experimental results show the feasibility of the proposed approach.

One-pass parallel thinning: analysis, properties, and quantitative evaluation

A one-pass parallel thinning algorithm based on a number of criteria, including connectivity, unit-width convergence, medial axis approximation, noise immunity, and efficiency, is proposed. A pipeline processing model is assumed for the development. Precise analysis of the thinning process is presented to show its properties, and proofs of skeletal connectivity and convergence are provided. The proposed algorithm is further extended to the derived-grid to attain an isotropic medial axis representation. A set of measures based on the desired properties of thinning is used for quantitative evaluation of various algorithms. Image reconstruction from connected skeletons is also discussed. Evaluation shows that the procedures compare favorably to others.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">&gt;</ETX>

Modelling a fast parallel thinning algorithm for shared memory SIMD computers

We present the mapping of a parallel thinning algorithm on shared memory SIMD computers. Four models of SM SIMD computers are considered. The parallel thinning algorithm is presented for all these four models of computers. The algorithm has been analyzed for time and cost performance on four models of SM SIMD computers. It has been shown that the CREW/CRCW SM SIMD thinning algorithm is cost optimal under a certain condition. A different condition makes the algorithm cost optimal on EREW/ERCW SM SIMD computers.

Optimization and performance analysis of thinning algorithms on parallel computers

This paper presents a concept for an implementation of different parallel thinning algorithms on parallel processors. The emphasis is put on a good parallelization using fine granularity and the simultaneous usage of vectorization. Results and experiences are discussed which have been obtained on the supercomputers CRAY X-MP/416 and CRAY Y-MP8/832 by realizing this concept, applied to one particular algorithm.

A new fast one-pass thinning algorithm and its parallel hardware implementation

A new parallel thinning algorithm and its parallel hardware implementation are proposed. The algorithm requires only a single pass of parallel operations per iteration, and a small number of iterations before obtaining satisfactory results. This makes the algorithm suitable for real-time image processing. The thinning results are also good in quality due to good selection of matching templates and the use of a two-level processing scheme.

A comparison of some one-pass parallel thinnings

A generalized expression for three existing one-pass parallel thinning algorithms is presented in this paper. It includes two fundamental terms, namely, thinning templates and a connectivity-preserving function. These two terms can be used to compare the efficiency and effects of these algorithms. Comparisons confirm that our algorithm is more efficient and effective than the other two.

Image compression algorithm for fast parallel thinning of binary images

An efficient image compression technique as a preprocessing step for skeletonisation is presented. For this purpose a 5 × 5 pattern window is used, in which an internal 3 × 3 nonoverlapping window does the compression whilst the outer ring of pixels helps to preserve the image connectivity during compression. Comparative evaluation based on experimental results, execution timings and quality of the skeleton obtained, show that this algorithm is faster and more efficient than many other existing thinning algorithms. This algorithm is also more efficient for noisy images than other thinning algorithms because it simultaneously performs median filtering operations.

Corrigendum to 'Parallel thinning with two-subiteration algorithms'

Corrigendum to 'Parallel thinning with two-subiteration algorithms'

A parallel thinning algorithm with fine grain subtasking

A one-pass iterative parallel thinning algorithm has been designed to reduce a two-dimensional pattern of strokes to its skeleton by removing layers of edge element until stroke has unitary thickness. An edge test computation is simulataneously carried out for each element in the two-dimensional pattern by inspecting the values of its neighbours. However, there is a degree of duplication among edge computations for elements with common neighbours. A modified algorithm described in this paper which avoids duplication at the cost of extra communication. The speed of the modified algorithm compared to the original algorithm is 1. (i) approximately 45% better on the ICL DAP; 2. (ii) approximately 30% better on the CRAY X-MP.

Parallel thinning with two-subiteration algorithms

Two parallel thinning algorithms are presented and evaluated in this article. The two algorithms use two-subiteration approaches: (1) alternatively deleting north and east and then south and west boundary pixels and (2) alternately applying a thinning operator to one of two subfields. Image connectivities are proven to be preserved and the algorithms' speed and medial curve thinness are compared to other two-subiteration approaches and a fully parallel approach. Both approaches produce very thin medial curves and the second achieves the fastest overall parallel thinning.