Research On The Indoor Location Algorithm Based On Vision

In recent years,with the expansion of buildings such as shopping malls,railway stations,exhibition hall,the demand for the indoor location services has become more urgent.Due to the satellite navigation system can not be used indoors,researchers start to look for other means to realize the indoor location service.At present indoor positioning technology include:infrared,Bluetooth,ultrasonic,WiFi,ultra wide band.The structure and environment of the building are different and complicated,besides there are different kinds of noises,as a result the precision and stability of the indoor positioning is low.For these reasons the application of indoor location service is limited.Indoor position technology based on visual became a hot spot which can provide high positioning accuracy,more scene information and convenient to carry.In order to increase the robustness and precision of localization algorithm,after summing of a lot of papers on indoor localization based on visual,the paper puts forward a new scheme include the scene recognition and classification process.The paper mainly studied the following two contents.(1)Using the convolution neural network to recognize and classify the indoor scenes.Because of the indoor scenes are complex and changeable,the traditional scene classification methods are gradually showing their limits.Convolution neural network can learn the features from a large number of images and show its great potential in the field of image classification.The paper use this method and achieve a high correct classification rate.(2)The paper put forward an improved algorithm which combines with the uncertainty of spatial reference points.Through the simulation experiment on simulated data and real scene,the proposed algorithm has been proven to be high accuracy and robustness.The results show that average error is 0.08 m,about 70% points positioning error is less than 0.1m,and about 90% of the points error is less than 0.2 m,maximum error is 0.3 m,reached the standard accuracy of indoor position.