Research And Implementation Of Content-based Image Search Engine

The image database is increasingly huge,gradually increase the demand of the corresponding image retrieval is imminent on the subject of exploration and research.From the large image database,how can in a very short time,quickly and efficiently to the needs of image retrieval and location,and has become the current urgent challenge topic.The image feature information in image retrieval is the most basic one,firstly,the related concept of image feature method are introduced,and then analyzes the traditional image features,the lack of support for the summary,make the corresponding optimization.And the color feature of the algorithm,the block color(gray)feature algorithm.Expounds the histogram algorithm,and introduces the co-occurrence matrix,the texture feature of Gabor filter,block Tarnura texture features,Tamura histogram,histogram two order moments algorithm,wavelet packet decomposition algorithm;algorithm of moment invariant shape feature.Secondly,based on the traditional image retrieval methods,the shortcomings and deficiencies are analyzed,the corresponding improvement measures,namely the retrieval method is more advanced,global feature fusion,and experiments are carried out and the corresponding computer simulation,results show that in image retrieval,the new method is more the high recall ratio and precision;in different categories of image retrieval,the global algorithm may have low discrimination feature fusion,aiming at this problem,this paper proposes characteristic feedback algorithm,its advantage is that the user can participate,through computer simulation,results show that by using this algorithm to search the whole After the system has been improved and optimized,its precision can be increased by 15 percentage points,and the recall rate will be increased by 10 percentage pointsFinally,the corresponding software is designed according to the above scheme.The content of the software is based on the point of view,and the image data is retrieved efficiently.At the same time,the advantages and disadvantages of the full text are summarized,and the future research directions are prospected...