- 1 0 年前最佳解答
In using the aspect in recent years, there is more and more obvious trend that shows the picture is searched, especially search on the basis of the image of the content of the picture and digit picture content service, have already become one door in the research field of information retrieval to study apparently.
Tradition deal with image method, picture of money search for the artificial characters, purpose even with person input key word that inquire about wanted, Search engine namely basis key words this carry on database than to stand side by side appear result of searching.
But the trend of dealing with the picture money of the image now has already become content search gradually, that is to say that the computer picks and fetches the picture money content characteristic value of the image as searching the target automatically.
Become riper even more with the development of image database, and the progress of the multimedia technology, add it in the borderless world of network, must have accumulated a large number of military picture money, so this research regards this as the goal, attempt to develop a set of military pictures money and search the method to classify in the information field of the national defence, try to develop the correlation diagram money in the field of military picture money and pick fetching and categorised method, the methods developed can offer the building and constructing of the follow-up military picture money database system.
2007-04-15 10:48:20 補充：
This research picks fetching and lamination characteristic to pick and fetch through the color characteristic, extract color of the primitive picture money and lamination characteristic value out, and classify the picture money after deal with leadingly for the transmitting type neural network,
2007-04-15 10:48:44 補充：
the main purpose of this research experiment is to distinguish the color characteristic and lamination characteristic,where under the circumstances that the transmitting type neural network trains,
2007-04-15 10:48:50 補充：
the automatic categorised success rate of the picture money is relatively high, in order to determine to be most suitable for the transmitting type characteristic value that classifies of military picture money of neural network