A new genetic-based technique for matching 3-D curves and surfaces

Different tasks in forensics require the use of 3D models of forensic objects (skulls, bones, corpses, etc.) captured by 3D range scanners. Since a whole object cannot be completely scanned in a single image using a range scanner, multiple acquisitions from different views are needed to supply the information to construct the 3D model by a range image registration method. There is an increasing interest in adopting evolutionary algorithms as the optimization technique for image registration methods. However, the image registration community tends to separate global and local searches in two different stages, named sequential hybridization approach, which is opposite to the scheme adopted by the memetic framework. In this work, we aim to analyze the capabilities of memetic algorithms (Moscato in On evolution, search, optimization, genetic algorithms and martial arts: towards memeticalgorithms. Report 826, Caltech Concurrent Computation Program, Pasadena, 1989) for tackling a really complex and challenging real-world problem as the 3D reconstruction of forensic objects. Our intention is threefold: firstly, designing new memetic-based methods for tackling a real-world problem and subsequently carrying out a performance and behavioral analysis of the results; secondly, comparing their performance with the one achieved by other methods based on the classical sequential hybridization approach; and thirdly, concluding the experimental study by highlighting the outcomes achieved by the best method in tackling the real-world problem. Several realworld 3D reconstruction problems from the Physical Anthropology Lab at the University of Granada, Spain, were used to support the evaluation study.

show abstract

“…We have used the Grid Closest Point (GCP) scheme (Yamany et al 1999) to speed up the computation of the closest point q cl of I m .…”

Section: Mathematical Formulationmentioning

confidence: 99%

“…2 Notice that we will not use the objective function of our previous contribution (Yamany et al 1999), to follow a more robust approach.…”

Section: Mathematical Formulationmentioning

confidence: 99%

Performance evaluation of memetic approaches in 3D reconstruction of forensic objects

et al. 2008

View full text Add to dashboard Cite

show abstract

“…− To search for the optimal geometric primitive correspondence in the matching space and then identify the appropriate transformation parameters-using numerical methods such as least square estimation-to overlay the scene and the model considering such matching (Besl and McKay, 1992;Zhang, 1994;Luck, Little, and Hoff, 2000;Liu, 2004), and − To directly search in the parameter space (usually by means of evolutionary algorithms), computing the matching between scene and model geometric primitives to validate the estimated transformation once it has been applied (Simunic and Loncaric, 1998;Yamany, Ahmed, and Farag, 1999;Han, Song, and Chung, 2001;He and Narayana, 2002;Cordón, Damas, and Santamaría, 2003).…”

Section: Image Registrationmentioning

confidence: 99%

“…In recent literature, we can find different approaches to matching and IR problems from the metaheuristics point of view, specially considering genetic algorithms (GAs) (Yamany, Ahmed, and Farag, 1999;Han, Song, and Chung, 2001). Such interest in applying evolutionary techniques to IR is related to the fact that the most natural representation for the problem is the real coding of the involved parameters and the good behavior of GAs in parameter learning is well known.…”

Section: Introductionmentioning

confidence: 99%

Image registration with iterated local search

Cordón

Damas

2006

J Heuristics

View full text Add to dashboard Cite

This contribution is devoted to the application of iterated local search to image registration, a very complex, real-world problem in the field of image processing. To do so, we first re-define this parameter estimation problem as a combinatorial optimization problem, then analyze the use of image-specific information to guide the search in the form of an heuristic function, and finally propose its solution by iterated local search.Our algorithm is tested by comparing its performance to that of two different baseline algorithms: iterative closest point, a well-known, image registration technique, a hybrid algorithm including the latter technique within a simulated annealing approach, a multi-start local search procedure, that allows us to check the influence of the search scheme considered in the problem solving, and a real coded genetic algorithm. Four different problem instances are tackled in the experimental study, resulting from two images and two transformations applied on them. Three parameter settings are analyzed in our approach in order to check three heuristic information scenarios where the heuristic is not used at all, is partially used or almost completely guides the search process, as well as two different number of iterations in the algorithms outer-inner loops.

show abstract

“…And a review in more recent works [3] [4] [5] [6] also reveals a tendency towards the employment of heuristic based search and optimization techniques, such as genetic algorithms, in order to improve image matching procedures for shape and object recognition tasks, with many advantages.…”

Section: Related Workmentioning

confidence: 99%

Object Detection for Computer Vision Using a Robust Genetic Algorithm

Centeno¹,

Lopes²,

Felisberto³

et al. 2005

Lecture Notes in Computer Science

View full text Add to dashboard Cite

Abstract. This work is concerned with the development and implementation of an image pattern recognition approach to support computational vision systems when it is necessary to automatically check the presence of specific objects on a scene, and, besides, to describe their position, orientation and scale. The developed methodology involves the use of a genetic algorithm to find known 2D object views in the image. The proposed approach is fast and presented a robust performance in several test instances including multiobject scenes, with or without partial occlusion.

show abstract

A new genetic-based technique for matching 3-D curves and surfaces

Cited by 79 publications

References 0 publications

Performance evaluation of memetic approaches in 3D reconstruction of forensic objects

Performance evaluation of memetic approaches in 3D reconstruction of forensic objects

Image registration with iterated local search

Object Detection for Computer Vision Using a Robust Genetic Algorithm

Contact Info

Product

Resources

About