Imputation of single-tree attributes using airborne laser scanning-based height, intensity, and alpha shape metrics

Forest inventories based on single-tree interpretation of airborne laser scanning (ALS) data often rely on an allometric estimation chain in which inaccuracies in the estimates of the diameter at breast height (DBH) propagate to other characteristics of interest such as the stem volume. Our purpose was to test nearest neighbor imputation by the k-Most Similar Neighbor (k-MSN) and the Random Forest (RF) methods for the simultaneous estimation of species, DBH, height and stem volume using ALS data. The predictors included computational alpha shape metrics and variables based on the height and intensity distributions in the ALS data. Separate data sets covering 1898 and 1249 dominant to intermediate trees in a typical Scandinavian stand structure were used for training and validation, respectively. RF proved to be a flexible method with an ability to handle 1846 predictors with no need for their reduction. Classification of Scots pine, Norway spruce and deciduous trees showed an accuracy of 78%, and the estimates of DBH, height and volume had root mean square errors of 13%, 3%, and 31%, respectively, when evaluated against the validation data. The two selection strategies implemented here reduced the number of candidate variables effectively without any substantial effect on the accuracy relative to the use of all predictors. Differences in k-MSN and RF imputations were marginal when the reduced sets of variables were used. Estimation accuracies could be maintained practically unchanged with only 12.5% of the initial reference data (237 trees), provided the distribution of the observations was similar in the reference and target data. Since we used information collected in the field for extracting the ALS point clouds for individual trees, our results represent an optimal case and should nevertheless be validated against automated tree delineation.     

Flexible view recognition for indoor navigation based on Gabor filters and support vector machines

Real-world scenes are hard to segment into (relevant) objects and (irrelevant) background. In this paper, we argue for view-based vision, which does not use segmentation, and demonstrate a practical approach for recognizing textured objects and scenes in office environments.

A small set of Gabor filters is used to preprocess texture combinations from input images. The impulse responses of the filters are transformed into feature vectors that are fed to support vector machines. Pairwise feature comparisons decide the classification of views.

We validate the approach on a robot platform using three different types of target objects and indoor scenes: people, doorways, and written signs. The general-purpose system can run in real time, and that recognition accuracies of practical utility are obtained.     

Classification consistency analysis for bootstrapping gene selection

Consistency modelling for gene selection is a new topic emerging from recent cancer bioinformatics research. The result of operations such as classification, clustering, or gene selection on a training set is often found to be very different from the same operations on a testing set, presenting a serious consistency problem. In practice, the inconsistency of microarray datasets prevents many typical gene selection methods working properly for cancer diagnosis and prognosis. In an attempt to deal with this problem, this paper proposes a new concept of classification consistency and applies it for microarray gene selection problem using a bootstrapping approach, with encouraging results.     

Modeling anisotropic light propagation in a realistic model of the human head

A Monte Carlo model capable of describing photon migration in arbitrary three-dimensional geometry with spatially varying optical properties and tissue anisotropy is presented. We use the model to explore the effects of anisotropy for optical measurements of the human head. An anisotropic diffusion equation that corresponds to our Monte Carlo model is derived, and a comparison between the Monte Carlo model and the diffusion equation solution with finite elements is given.     

MORE – a multimodal observation and analysis system for social interaction research

The MORE system is designed for observation and machine-aided analysis of social interaction in real life situations, such as classroom teaching scenarios and business meetings. The system utilizes a multichannel approach to collect data whereby multiple streams of data in a number of different modalities are obtained from each situation. Typically the system collects a 360-degree video and audio feed from multiple microphones set up in the space. The system includes an advanced server backend component that is capable of performing video processing, feature extraction and archiving operations on behalf of the user. The feature extraction services form a key part of the system and rely on advanced signal analysis techniques, such as speech processing, motion activity detection and facial expression recognition in order to speed up the analysis of large data sets. The provided web interface weaves the multiple streams of information together, utilizes the extracted features as metadata on the audio and video data and lets the user dive into analyzing the recorded events. The objective of the system is to facilitate easy navigation of multimodal data and enable the analysis of the recorded situations for the purposes of, for example, behavioral studies, teacher training and business development. A further unique feature of the system is its low setup overhead and high portability as the lightest MORE setup only requires a laptop computer and the selected set of sensors on site.     

High levels of equol in organic skimmed Finnish cow milk

The isoflavonoids, equol, formononetin, daidzein, genistein, biochanin A, and O-demethylangolensin (O-DMA), were analyzed from commercial cartons of skimmed Finnish milk by HPLC-diode array detector (DAD)-FL. We found 411 +/- 65 ng/mL of equol and traces of formononetin and daidzein in organic skimmed milk whereas conventionally produced milk contained 62 +/- 16 ng/mL of equol and no formononetin or daidzein.     

On simplification of large fault trees

This paper describes some fault tree optimization algorithms used in STUK PSA code SPSA that is capable of generating minimal cut sets for fault trees containing 12 000 gates and basic events. The complexity of a minimal cut set search is a function of multiple gates and basic events. In a fault tree containing 1000 multiple elements, these optimizations reduce the number of multiple elements often by 30 to 50%. One novel feature of the algorithm is that no boolean reduction rules are used, since the fault tree itself contains its simplification rules.