Object detection and automatic measurements with a data-driven algorithm for marked point process optimization
| Autor: | Coiffard Marre, Claire, Geffray, Ségolen |
|---|---|
| Přispěvatelé: | Institut de Recherche Mathématique Avancée (IRMA), Université de Strasbourg (UNISTRA)-Centre National de la Recherche Scientifique (CNRS), Coiffard Marre, Claire |
| Jazyk: | angličtina |
| Rok vydání: | 2014 |
| Předmět: |
[MATH.MATH-PR] Mathematics [math]/Probability [math.PR]
maximum a posteriori [STAT.TH] Statistics [stat]/Statistics Theory [stat.TH] [INFO.INFO-TS] Computer Science [cs]/Signal and Image Processing [STAT.TH]Statistics [stat]/Statistics Theory [stat.TH] Simulated annealing [MATH.MATH-PR]Mathematics [math]/Probability [math.PR] [INFO.INFO-TS]Computer Science [cs]/Signal and Image Processing reversible jump Markov Chain Monte-Carlo circular Hough transform [MATH.MATH-ST]Mathematics [math]/Statistics [math.ST] Spatial marked point processes [MATH.MATH-ST] Mathematics [math]/Statistics [math.ST] [SPI.SIGNAL]Engineering Sciences [physics]/Signal and Image processing [SPI.SIGNAL] Engineering Sciences [physics]/Signal and Image processing Radon transform |
| Popis: | 29 pages; This paper deals with object detection and automatic measurements from optical microscopy and scanning electron microscopy (SEM) 2-dimensional images. More specifically, we consider the problem of detecting and measuring circular cells and rectilinear thick fibers. We adopt the spatial point process viewpoint and follow the maximum a posteriori (MAP) principle to obtain an optimal object configuration. We design a novel data-driven version of the data term in the MAP criterion in the fiber case, of the reference spatial marked point process and of the Markov Chain Monte-Carlo (MCMC) sampler, both in the cell case and in the fiber case. To this purpose, information from the image under consideration is incorporated by means of either the Radon transform or the circular Hough transform. We apply the proposed method to a real cell optical microscopy image and to a real fiber SEM image. Our results show that the proposed algorithm is fast, reliable and stable. We also produce the required characterization of the object morphology. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|