Research On Statistical Algorithm Of Staying Tourists In Garden Scenic Spot

In order to explore the inherent law of the ingenious combination of the path space,the research of resident tourists can be used as a breakthrough point to carry out the quantitative analysis of the broad and profound Chinese classical garden.The number of daily visitors in the garden is huge,and their distribution situation is complex in the path.It is unrealistic to use the traditional method to count the number of tourists.In the long run,to repeat visitor detecting and improve the science,accuracy and efficiency of the statistical data,the machine vision technology was applied to get the number of tourists in different periods of time by the way of image processing and pattern recognition method.The data was supporting for the analysis of classical garden design and the inherent behavior of tourists.The Master of Nets Garden is considered as the research object in this paper.By field investigation and research,we designed and built up the Master of Nets Garden video monitoring system to gather the real time data of stationary points.We studied various foreground detection algorithms to extract the main characteristics of the target.We analyzed various tracking algorithms to realize the accurate tracking of multi-targets.Then,we designed the appropriate resident counting method,developed special software,and completed the core tasks of the project:Accurate statistics on the resident volume of tourists in the garden area.The main research contents can be summarized as follows:(1)According to the different stagnation situation investigated in the Master of Nets Garden,we designed and completed the entire network division video acquisition system,consisting of equipment selection,installation and commissioning,and video admission storage.For each quarter,we made a week of video acquisition,and completed the stitching and format conversion of the video.(2)Considering the complex scenes of lighting and shaking motions,swinging leaves and ripples in the garden,a new ViBe algorithm based on Lab color space was proposed by deep research on various target foreground detection algorithms.The algorithm had a better effect in eliminating the shadow and improved the accuracy of foreground detection.Then we realized the accurate feature extraction of the main features such as target histogram,texture,position and area in the video.(3)We studied the current tracking algorithm,and showed the single target tracking effect of Mean Shift and CamShift.In order to solve the multi-target tracking problem,an improved Mean Shift multi-target tracking algorithm was proposed in this paper.And we studied the multi-target tracking algorithm based on color histogram.Comparing several tracking algorithms,we thought out the most relevant algorithm for the subject.(4)MFC and OpenCV library were utilized In Visual Studio2013 environment to develop resident statistics software,which realized these functions,such as parameter settings,location display,video progress,team display and so on.It also put 10 kinds of resident in different periods of tourists' data into the foreground Excel spreadsheet every five minutes(The intervals are 5s,6s,8s,10s,12s,20s,40s,60s,90s,150s).(5)We studied the data,tested software performance and efficiency,analyzed the software accuracy.After the test analysis,the software running speed and video real-time playback synchronization,a computer could run multiple instances at the same time,greatly reducing the data acquisition time.We tested the influence of the space type,light intensity and traffic volume on the accuracy of the data.Each quarter we selected one day to count stay for 5 second's tourists.With the comparison of data calculated by human and program,we got the result that the accuracy rate of average data scored 82.72%.We took the factors that affect the accuracy of the data into consideration for later optimization.