基于图像匹配的高精度室内定位技术研究

Indoor Positioning Technology Based on Image Matching with High Accuracy

本文提出了一种新的室内定位方法,上传由手机摄像头拍摄的照片,在定位图像库中进行场景选择、图像匹配,得到最佳匹配图像,根据该图像的预留信息实现准确的室内定位。首先,利用深度卷积神经网络特征和无监督学习方法进行场景选择,然后采用ASIFT(Affine Scale Invariant Feature Transform)算法对既定场景的图像逐一匹配,同时加入对极几何约束,进一步提高定位精度和速度。这种基于图像匹配的室内定位技术能够同时得到用户的位置信息和拍照时的方向信息。

This paper puts forward a new method of indoor positioning,upload the photographs taken by cell phone cameras,in positioning image library for scenario selection,image matching,image to get the best match,according to the image of the reserved information to achieve accurate indoor positioning.First of all,Using the feature of the deep convolution neural network and the unsupervised learning method to select the scene,And then apply ASIFT（Affine Scale Invariant Feature Transform）The algorithm matches the image of the established scene one by one,while adding the geometric constraints,further improving the positioning accuracy and speed.This image matching based indoor positioning technology can simultaneously obtain the user’s location information and the direction information of the photograph.