Research On The Method Of Sparse Representation For Typical Indoor Scenes

At the end of 90 s of last century,the research of image scene classification,With the development and popularization of information technology and intelligent robot,as one of the important research contents,scene recognition has become a key research topic in the field of computer vision and pattern recognition.As one of the most important research fields in image scene classification,indoor scene classification is one of the most important research topics.If we can effectively improve the accuracy of indoor scene classification,indoor scene classification will play a greater role in image retrieval,intelligent robots and other related fields.Therefore,the study of indoor scene classification is important and meaningful.In this paper,we propose a new method of indoor scene classification based on image saliency detection and sparse representation,in order to reduce the rate of correct recognition in the scene recognition.The system is based on the significant detection of the image of the home indoor scene,the main target of the image obtained by the sparse representation of the results to achieve the classification of indoor scenes.The main contents of this paper are as follows: the research status and significance of the current scene image and indoor scene image are studied.Introduces the basic principle of sparse representation and some common algorithms,mainly analyzes the dictionary learning and sparse solution of the two aspects of the algorithm,such as MOD method,OMP algorithm,and choose the most suitable for indoor scenes.In order to improve the classification accuracy of indoor scene images,this paper introduces the method of saliency detection.Through the saliency detection of the indoor scene image,we can get the main target image of this kind of scene,and then realize the classification of the indoor scene image by the sparse recognition of the significant main target image.Some common saliency detection methods are studied,and algorithm which is more suitable for indoor scene image.And the image feature of the saliency detection is combined with other features of the image,so as to further improve the classification effect of the indoor scene image.Comparative experiment was designed with other classification,the results and data was analyzed shows that our system can improve the classification accuracy to a certain extent.