Registration of 3-D images by genetic optimization

Automatic Registration of TLS-TLS and TLS-MLS Point Clouds Using a Genetic Algorithm

Registration of point clouds is a fundamental issue in Light Detection and Ranging (LiDAR) remote sensing because point clouds scanned from multiple scan stations or by different platforms need to be transformed to a uniform coordinate reference frame. This paper proposes an efficient registration method based on genetic algorithm (GA) for automatic alignment of two terrestrial LiDAR scanning (TLS) point clouds (TLS-TLS point clouds) and alignment between TLS and mobile LiDAR scanning (MLS) point clouds (TLS-MLS point clouds). The scanning station position acquired by the TLS built-in GPS and the quasi-horizontal orientation of the LiDAR sensor in data acquisition are used as constraints to narrow the search space in GA. A new fitness function to evaluate the solutions for GA, named as Normalized Sum of Matching Scores, is proposed for accurate registration. Our method is divided into five steps: selection of matching points, initialization of population, transformation of matching points, calculation of fitness values, and genetic operation. The method is verified using a TLS-TLS data set and a TLS-MLS data set. The experimental results indicate that the RMSE of registration of TLS-TLS point clouds is 3~5 mm, and that of TLS-MLS point clouds is 2~4 cm. The registration integrating the existing well-known ICP with GA is further proposed to accelerate the optimization and its optimizing time decreases by about 50%.

Genetic algorithms for a robust 3-D MR-CT registration

The aim of this paper is to present an original usage of genetic algorithms as a robust search space sampler in application to 3-D medical image elastic registration. An overview of the standard steps of a registration algorithm is given. We focus on the genetic algorithms use and particularly on the problem of extraction of the optimal solution among the final genetic population. We provide an original encoding scheme relying on a structural approach of point matching and then point out the need for a local optimization process. We then illustrate the algorithm with a concrete registration example and assert the results with a direct multivolume rendering tool. Finally, the algorithm is applied on the vanderbilt medical image database to assert the robustness and in order to compare it with other techniques.

Automatic realignment of time-separated MR images by genetic algorithm

A simple yet highly efficient artificial intelligence technique utilizing a genetic algorithm is used to register time-separated pairs of MRI data sets. To encourage others to try the approach, the algorithm is presented by way of a simple example to a 2-D data set; it is equally applicable to 3-D data. The technique is reliably found to reduce mismatch in images of the distal-interphalangeal joint from the order of several mm to just 200 microm (one pixel). The method and transformation are general and would be suitable for locating physical changes between any image data sets. We believe the technique to be of use in functional imaging, measurement of disease progression with time (e.g., degradation of cartilage in arthritic disease) and pre/post-surgical studies.

A direct multi-volume rendering method aiming at comparisons of 3-D images and models

We present a new method for direct volume rendering of multiple three-dimensional (3-D) functions using a density emitter model. This work aims at obtaining visual assessment of the results of a 3-D image registration algorithm which operates on anisotropic and non segmented medical data. We first discuss the fundamentals associated with direct, simultaneous rendering of such datasets. Then, we recall the fuzzy classification and fuzzy surface rendering theory within the density emitter model terminology, and propose an extension of standard direct volume rendering that can handle the rendering of two or more 3-D functions; this consists of the definition of merging rules that are applied on emitter clouds. The included rendering applications are related on one hand, to volume-to-volume registration, and on the other hand, to surface-to-volume registration: the first case is concerned with global elastic registration of CT data, and the second one presents fitting of an implicit surface over a CT data subset. In these two medical imaging application cases, our rendering scheme offers a comprehensive appreciation of the relative position of structural information.