Detection Of Moving Objects In Water Environment Based On Video Surveillance System

With the influence of global warming and human activities,the water environment around the world has changed to a certain extent,manifested in the increase in floating objects such as garbage and algae in the water environment(including lakes and offshore waters)and the water quality in many areas is declining.Therefore,it is necessary and important to monitor the water environment in many aspects.This thesis starts from the description of the defects of existing measurement methods for moving targets in the water environment,and present innovative methods based on computers and target detection algorithms to improve the monitoring of moving targets in aquatic environment.Specifically,two targets are aimed here:one is the measurement of water transparency via Secchi disk in water,the other is floating materials on water surface.Water transparency is one of the basic parameters of water quality.For more than 100 years,people have measured this property through observation of a disk(Secchi disk)lowered in water that is connected with a rope.When the human eye cannot see the Secchi disk,the depth of the disk in water is defined as the transparency of the water body,also known as Secchi disk depth(ZSD).The measurement results of this method depend entirely on human judgment,lacking objectivity and stability.This study proposes a method to determine the ZSD by replacing the human eye with a digital camera.Specifically,an improved background subtraction algorithm is used to distinguish the Secchi disk from other background objects.The time when the Secchi disk is indiscernible from all background pixels are regarded as the corresponding ZSD time.The experimental results show that,compared with the traditional method,the uncertainty of the ZSD measured by the new system reaches 3.7%,while the uncertainty of the traditional human eye observation method is 11.9%.The results show that the new electronic system is not only objective but also more stable,reducing the subjectivity brought about by human eye discrimination.In addition,we have also designed and manufactured a lightweight and stable measurement system,so that the Secchi disk remains vertical when the water flow is very fast,so that accurate measurements can be made in a water environment of the strong current.On the other hand,water environments of tourist attractions and intake cooling water for nuclear power plants,a large number of floating objects often appear,including algae,Acaudina molpadioides,shrimps or garbage,etc.,which not only damage the ecological environment of the water,but also may block the filtration system of the nuclear power plant,affecting its normal operation and causing large economic losses.It is necessary to monitor these floating objects in real time.However,the current real-time monitoring system for long-range floating objects has a few shortcomings.For example,the detection capabilities of ordinary commercial cameras for detecting floating objects are not strong enough,and they are easily affected by environmental factors(such as water surface reflection)during the monitoring process,and lack of location information of the target.In order to improve the detection and positioning capabilities of the monitoring system,it is necessary to modify the commonly used camera lenses.To this end,this study explored two aspects:1)adding a polarizer in front of the lens to reduce the impact of water reflection;2)modifying the camera lens image acquisition device to enhance its detection capabilities.The results show that the kappa coefficient of the image acquisition device after modification is increased from 0.70 to 0.90,which shows significantly improved monitoring capabilities.In addition,an algorithm based on the geometric model of the camera image is used to establish the position information of floating objects in the image.These results not only provide more effective monitoring of the water environment,but also lay the foundation for predicting the future position of floating objects through dynamic models.