Research On Fuzzy Retrieval Method Of Civil Aviation Checked Baggage Based On Semantics

Image retrieval based on semantic technology is a hot spot in computer vision research.Aiming at the application of semantic retrieval in the field of civil aviation,this paper studied the method of retrieving the lost baggage of passengers.Firstly,a feature ranking method based on saliency was proposed,and then the concept of user relevance feedback was added to improve the accuracy of baggage image retrieval.The main research work is as follows:Firstly,in order to solve the problem that the dim background of the baggage image has a bad effect on the retrieval results in image retrieval,an image segmentation algorithm based on the combination of improved watershed and fuzzy mean was proposed.The oversegmentation of watershed is solved by the multi-scale morphological gradient reconstruction method,which can preserve the object contour details and eliminate the noise and useless gradient details.Then the image was segmented twice by the fuzzy mean method,and the experiment proves that the method was suitable for most of the luggage images.Secondly,in order to improve the retrieval accuracy and shorten the retrieval time,the application of reference significance in the image,as well as the formation of the algorithm of analyzing three features,the feature saliency was defined.By calculating the saliency of CCV color,sift and LDP texture features,the significance of the three features in the image during the retrieval process was determined.The strategy of feature priority retrieval with high saliency was adopted to improve the overall retrieval performance.Experiments show that the definition of saliency was effective,and the retrieval efficiency was improved by this method.Finally,in order to better reflect the user's demand for retrieval results,the concept of semantic based relevance feedback was added.Firstly,the user makes subjective evaluation on the results of the first round of retrieval through the retrieval interface,and then the system redetermines the significance of the three features,so as to retrieve the retrieval results more in line with the user's needs.The experiment proves that the method can achieve the expected effect.