Tratamiento de imágenes acuáticas para el estudio del fondo marino
| Autor: | Benito Martı́nez, Joan |
|---|---|
| Přispěvatelé: | Pujol Vilanova, Núria |
| Jazyk: | Catalan; Valencian |
| Rok vydání: | 2017 |
| Předmět: | |
| Zdroj: | Dipòsit Digital de la UB Universidad de Barcelona |
| Popis: | Treballs Finals de Grau d'Enginyeria Informàtica, Facultat de Matemàtiques, Universitat de Barcelona, Any: 2017, Director: Núria Pujol Vilanova [en] The final project of my degree, Treatment of aquatic images for the study of the seafloor, talks about the search of problems that we can have to take aquatic photographies. Also, we can know the possible ways to determine and improve the results obtained. The images in this project are taken from a sporty camera anchored to a robot ship that is unmanned. This ship tours a programmed way in a region, always two meters from the seafloor. However, the results obtained for the camera aren’t professionals, so the main objective is to find the best way to treat the images by a software and try to get valid images to do any study with them. Histogram equalization should be the answer, I have been able to check that adaptative equalization with algorithms CLAHE is a good option. Furthermore, these images suffer the movement of the ship. The movement is caused by the addition of underwater currrents, the vibration of the engines and the auto correcction in order to follow the way. I have wanted to demostrate this event creating a kind of panoramic. The purpose is to see how far the ship suffers the deviation. I have proposed a way for the deviation calculation. This way, consist in compare centers of the image with the projection and translation that it suffers while the panoramic was done. So that, the calculation give us a value pretty close comparing with the maker. It’s approaching to 3% of error rate. Finally, with the little remaining time, I have introduced a little purpose to detection of vegetation. It’s an orientative way in order to search pixels inside a colour range. This method can improve with cascade classifiers or with the characteristics to indentify pattern. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|