False alarms in home surveillance systems are an annoyance for customers and it has become a much-debated topic. There are several previous studies where the purpose is to reduce the number of false alarms in home surveillance systems, but not about combining different types of sensors in different sensitivities that complement each other.The purpose of the thesis is to combine three motion detectors; two passive infrared (PIR) sensors and a microwave sensor coupled to an Arduino Uno with different sensitivities and compare it with each individual motion detector with different sensitivities, to determine what gives the best results in detecting people for an indoor monitoring system. An experiment with these sensors has been carried out and the data from the sensors is trained in a machine learning function which then plotted a ROC graph with a curve summarizing the result.PIR-sensors are triggered by heat and detect human thermal energy. Combined with a microwave sensor that senses human movements, the number of false alarms was reduced with an area under curve (AUC) of up to 97% with the lowest sensitivity of the sensors.The method that has been used can be reused for further studies for tests with other sensors and combinations. The method selection with machine learning function that trains data from the sensors and draws out a ROC graph gave a very clear picture of how well that particular combination of sensors and sensitivity works. By interpreting the ROC curve, combining several sensors has shown to give better results. The PIR-sensor has shown to give a low false positive rate by reducing the number of false alarms when no person passes, the microwave sensor has shown to give a high true positive rate, by reducing the number of misses when a person passes. The combination of the sensors has led to the sensors complementing each other which reduced the number of false alarms.
| Identifer | oai:union.ndltd.org:UPSALLA1/oai:DiVA.org:hj-48065 |
| Date | January 2020 |
| Creators | Ismail Ahmed, Makail, Sürer, Antonio |
| Publisher | Tekniska Högskolan, Jönköping University, JTH, Datateknik och informatik, Tekniska Högskolan, Jönköping University, JTH, Datateknik och informatik |
| Source Sets | DiVA Archive at Upsalla University |
| Language | Swedish |
| Detected Language | English |
| Type | Student thesis, info:eu-repo/semantics/bachelorThesis, text |
| Format | application/pdf |
| Rights | info:eu-repo/semantics/openAccess |
Page generated in 0.0016 seconds