| dc.description.abstract | In this paper, we present one important aspect of a complete application of numerical image analysis in biology. We want to analyse a biological object through serial cuttings to measure morphological and biochemical parameters. To achieve this, we have to deal with the problem of distortion which appears between the serial cuttings, coming from the biological objects themselves or from the technology used in experimentation process. Our objective is to build a robust routine application. Therefore we have made an effort on the automation degree and the quality of the generated measures. We propose a solution based on a global polynomial transformation model to take into account the distortions and a relaxation process to estimate the model coefficients. The association of these two steps gives a new tool which opens new perspectives to the biologist using even now qualitative or manually obtained quantitative informations to study the biological objects. | en_US |
