Name: | Description: | Size: | Format: | |
---|---|---|---|---|
19.71 MB | Adobe PDF |
Authors
Advisor(s)
Abstract(s)
A Ilha da Madeira ao longos dos anos tem sido fustigada pela ocorrência de aluviões
com consequências por vezes prejudiciais para a população. Devido a esta situação foi
possível concluir que era necessário um sistema de monitorização do nível de água nas
ribeiras para monitorização de situações de risco de aluviões.
Para a monitorização do nível de água em rios, ribeiras ou cursos de água urbanos
é considerado cada vez mais procedimentos de processamento de imagem. A maioria
dos sistemas utiliza medidores de água, que são réguas que auxiliam a medição do
nível de água. No entanto, estes sistemas podem falhar, quando são aplicados a cursos
de água urbanos, devido à ondulação da água, detritos existentes na superfície da água,
a traços de chuva ou outros efeitos adversos capturados nas imagens que podem
provocar resultados erróneos. A importância de considerar todos estes efeitos é que
estes geralmente estão associados à variação do nível de água com a ocorrência de
precipitação.
O sistema desenvolvido neste projeto contém uma câmara infravermelha
posicionada num prédio a apontar para a ribeira de Santa Luzia, no concelho do
Funchal. Esta câmara, pelas suas características, permite operar durante o dia e durante
a noite. A zona de captura da imagem é maior do que sistemas existentes para poder
minimizar os efeitos que tendem a obstruir a linha de água.
Inicialmente foi desenvolvido um sistema de referência para que fosse possível
converter os pixéis numa unidade de medição, neste caso, em centímetros. De seguida
procedeu-se a implementação do algoritmo de forma a obter o nível de água. Primeiro
aplicaram-se certas técnicas de forma a melhor a qualidade da imagem. Depois efetuou se a compensação do movimento da câmara através da deteção de uma das bordas
superiores da ribeira para compensar a rotação da câmara e através da deteção de
padrões para compensar a translação da câmara. De seguida foi obtida a região de
interesse e analisou-se a mesma em diferentes condições climatéricas por forma a obter
a linha de água para cada situação. Neste processo foi efetuada uma média temporal e
espacial de imagens para obter um melhor perfil de escala de cinza e de gradiente de
cada imagem. Por fim foi obtido o nível de água através do sistema de referência.
A exatidão teve uma variação entre 0,7 cm e 1,6 cm durante o dia e entre 2,7 cm e
3,3 cm durante a noite. A exatidão média obtida através das diferentes condições
climatéricas foi de 1,1 cm durante o dia e 2,9 cm durante a noite. Uma das soluções
para melhorar a exatidão da deteção do nível de água durante a noite é através da
implementação de iluminação artificial orientada para a região de interesse.
A técnica desenvolvida permite obter uma boa exatidão para diferentes condições
climatéricas, combinando informações de várias posições de deteção da linha de água
para evitar problemas na zona de deteção. De referir que deste trabalho resultou um
artigo publicado em revista.
Madeira Island over the years has been fustigate by the occurrence of flooding with sometimes harmful consequences for the population. Due to this situation, it was possible to conclude that a system to monitor the water level in the streams was necessary to control alluvium. For water level monitoring in rivers, streams or urban streams, image processing procedures are increasingly being considered. Most systems use staff gauges, which are rulers that assist in measuring the water level. However, these systems can fail when applied to urban waterways due to water ripples, debris on the water surface, traces of rain, or other adverse effects captured in the images that can cause erroneous results. The importance of considering all these effects is that they are generally associated with the variation of the water level with the occurrence of precipitation. The developed system in this project contains an infrared camera positioned on a building to the Santa Luzia stream, in the municipality of Funchal. This camera, due to its characteristics, can operate during day and night. The image capture zone is larger than existing systems to minimize the effects that tend to obstruct the water line. Initially, a reference system was developed so that it was possible to convert the pixels into a measurement unit, in this case centimeters. Then we proceeded to implement the algorithm to obtain the water level. First, some techniques were performed to improve the image quality. Then the camera movement was compensated by detecting one of the upper edges of the stream to compensate for camera rotation and by detecting patterns to compensate for camera translation. Then the region of interest was obtained and analyzed in different weather conditions to obtain the waterline for each situation. In this process, temporal and spatial averaging of images was performed to obtain a better grayscale and gradient profile for each image. Finally, the water level was obtained through the reference system. The accuracy ranged from 0.7 cm to 1.6 cm during the day and from 2.7 cm to 3.3 cm at night. The average accuracy obtained across the different weather conditions was 1.1 cm during the day and 2.9 cm at night. One of the solutions to improve the accuracy of water level detection at night is through the implementation of artificial lighting oriented to the region of interest. The developed technique allows to obtain good accuracy for different weather conditions by combining information from several waterline detection positions to avoid problems in the detection zone. If should be noted that this work resulted in an article published in a journal.
Madeira Island over the years has been fustigate by the occurrence of flooding with sometimes harmful consequences for the population. Due to this situation, it was possible to conclude that a system to monitor the water level in the streams was necessary to control alluvium. For water level monitoring in rivers, streams or urban streams, image processing procedures are increasingly being considered. Most systems use staff gauges, which are rulers that assist in measuring the water level. However, these systems can fail when applied to urban waterways due to water ripples, debris on the water surface, traces of rain, or other adverse effects captured in the images that can cause erroneous results. The importance of considering all these effects is that they are generally associated with the variation of the water level with the occurrence of precipitation. The developed system in this project contains an infrared camera positioned on a building to the Santa Luzia stream, in the municipality of Funchal. This camera, due to its characteristics, can operate during day and night. The image capture zone is larger than existing systems to minimize the effects that tend to obstruct the water line. Initially, a reference system was developed so that it was possible to convert the pixels into a measurement unit, in this case centimeters. Then we proceeded to implement the algorithm to obtain the water level. First, some techniques were performed to improve the image quality. Then the camera movement was compensated by detecting one of the upper edges of the stream to compensate for camera rotation and by detecting patterns to compensate for camera translation. Then the region of interest was obtained and analyzed in different weather conditions to obtain the waterline for each situation. In this process, temporal and spatial averaging of images was performed to obtain a better grayscale and gradient profile for each image. Finally, the water level was obtained through the reference system. The accuracy ranged from 0.7 cm to 1.6 cm during the day and from 2.7 cm to 3.3 cm at night. The average accuracy obtained across the different weather conditions was 1.1 cm during the day and 2.9 cm at night. One of the solutions to improve the accuracy of water level detection at night is through the implementation of artificial lighting oriented to the region of interest. The developed technique allows to obtain good accuracy for different weather conditions by combining information from several waterline detection positions to avoid problems in the detection zone. If should be noted that this work resulted in an article published in a journal.
Description
Keywords
Aluvião Monitorização Cursos de água Medição do nível de água Processamento de imagem Funchal, Madeira (Portugal) Alluvium Monitoring Streams Water level measurement Image processing Engenharia Eletrotécnica-Telecomunicações . Faculdade de Ciências Exatas e da Engenharia