Research On Position Recognition Based On Vision

Visual position recognition is one of the most basic topics in the field of computer vision.Its task is to accurately and effectively identify positions based on a given query image.For a long time,due to the increasing focus on robotic autonomy and the rapid decline in the cost of visual sensors,the research of visual position recognition has received extensive attention and rapid development.So far,there are still a series of challenges in the field of visual position recognition.One of the challenges is that multiple locations in the environment look very similar,and the same location will look different from different perspectives,which can lead to the problem of confusion in perception.In response to the above problems,this paper proposes a new algorithm to provide a solution to the problem.Spatial information is segmented by constructing classification labels of locations to avoid the problem of perceived confusion,and at the same time it reduces the search space and improves efficiency.This method uses a saliency algorithm for the corresponding video sequence segment that belonging the each locationing key frames in the position recognition system to generate a large number of visually significant candidate objects,and effectively calculates the evaluation function between these candidate objects,then The hierarchical clustering algorithm is used to calculate the representative objects in each sequence,finally these objects are combined into a sign with a representative video sequence.Using the way of signage to represent the place,insert it in the previous step of the large number of image sets corresponding to the search place in the position recognition system,in order to narrow the search range and avoid the confusion of global search uncertainty caused by perceived confusion.The main research results of this paper are: Investigating the development of visual position recognition,researching and analyzing algorithms of great significance in the research process of visual position recognition,and proposing new solutions to the problems that arise in it.At the same time,it investigates the development of saliency detection algorithms,and comparatively analyzes algorithms that are of great significance in the research process of saliency detection algorithms.Finally,for the problem of perceived confusion,a variety of signage methods to solve this problem are proposed.