Ragnar Bang Huseby

Applications of hidden Markov chains in image analysis

In image analysis, two-dimensional Markov models, i.e. Markov field models, have been applied for segmentation purposes, but except for the area of text recognition, the application of hidden Markov chains has been rare. Through four very different examples, this paper demonstrates the applicability also for hidden Markov chains in image analysis, and shows that the problems of image analysis often may have one- dimensional characteristics even though the images are two-dimensional.

Space–time modelling of the spread of pancreas disease (PD) within and between Norwegian marine salmonid farms

Infectious diseases are a constant threat to industrialised farming, which is characterised by high densities of farms and farm animals. Several mathematical and statistical models on spatio-temporal dynamics of infectious diseases in various farmed host populations have been developed during the last decades. Here we present a spatio-temporal stochastic model for the spread of a disease between and within aquaculture farms. The spread between farms is divided into several transmission pathways, including (i) distance related spread and (ii) other types of contagious contacts. The within-farm infection dynamics is modelled by a susceptible-infected-recovered (SIR) model. We apply this framework to model the spread of pancreas disease (PD) in salmon farming, using data covering all farms producing salmonids over 9 years in Norway. The motivation for the study was partly to unravel the spatio-temporal dynamics of PD in salmon farming and partly to use the model for scenario simulation of PD control strategies. We find, for example, that within-farm infection dynamics vary with season and we provide estimates of the timing from unobserved infection events to disease outbreaks on farms are detected. The simulations suggest that if a strategy involving culling of infectious cohorts is implemented, the number of detected disease outbreaks per year may be reduced by 57% after the full effect has been reached. We argue that the high detail and coverage of data on salmonid production and disease occurrence should encourage the use of simulation modelling as a means of testing effects of extensive control measures before they are implemented in the salmon farming industry.

Adaptive Registration of Remote Sensing Images using Supervised Learning

This paper describes a system for co-registration of time series satellite images which uses a learning-based strategy. During a training phase, the system learns to recognize regions in an image suited for registration. It also learns the relationship between image characteristics and registration performance for a set of different registration algorithms. This enables intelligent selection of an appropriate registration algorithm for each region in the image, while regions unsuited for registration can be discarded. The approach is intended for co-registration of sequences of images acquired from identical or similar earth observation sensors. It has been tested for such sequences from different types of sensors, both optical and radar, with varying resolution. For images with moderate differences in content, the registration accuracy is, in general, good with an RMS error of one pixel or less.

A stage-structured Bayesian hierarchical model for salmon lice populations at individual salmon farms – Estimated from multiple farm data sets

Salmon farming has become a prosperous international industry over the last decades. Along with growth in the production farmed salmon, however, an increasing threat by pathogens has emerged. Of special concern is the propagation and spread of the salmon louse, Lepeophtheirus salmonis. To gain insight into this parasites population dynamics in large scale salmon farming system, we present a fully mechanistic stage-structured population model for the salmon louse, also allowing for complexities involved in the hierarchical structure of full scale salmon farming. The model estimates parameters controlling a wide range of processes, including temperature dependent demographic rates, fish size and abundance effects on louse transmission rates, effect sizes of various salmon louse control measures, and distance based between farm transmission rates. Model parameters were estimated from data including 32 salmon farms, except the last production months for five farms, which were used to evaluate model predictions. We used a Bayesian estimation approach, combining the prior distributions and the data likelihood into a joint posterior distribution for all model parameters. The model generated expected values that fitted the observed infection levels of the chalimus, adult female and other mobile stages of salmon lice, reasonably well. Predictions for the periods not used for fitting the model were also consistent with the observational data. We argue that the present model for the population dynamics of the salmon louse in aquaculture farm systems may contribute to resolve the complexity of processes that drive this host-parasite relationship, and hence may improve strategies to control the parasite in this production system.