The Theoretical Research In Texture Video Searching Of The Main Tree's Panel Characteristic In The Northeast

At present, the price of many precious varieties of trees material is ten times higher than the ordinary log, so that it has sizable commercial value to recognize the kind of panel material. Because the artificial expert has lower relatively perceptual knowledge to material of these precious varieties of trees, it have increased the difficulty to the artificial vision discriminating. Some illegal retailers mix the spurious with the genuine, adulterate, upset the market of panel, so, It will drive whole panel texture recognition scientific research field to set up of our country panel macroscopically characteristic CD retrieval system and research panel texture recognition technology, and can fully utilize our country's existing timber resource advantage and utilize and import resources better.The texture is the important characteristic of lumber, the different kind of tree's texture has the certain difference. This article takes the main kind of trees in the northeast area in our country as research objects, through gathering panel texture image, removes its background, obtains image only with the target texture, and make the color image gradation, carries on median filter, eliminates the disturbing noise in the image, carries on two values processing and detecting brim;Analysis textures characteristic of several kind of main trees, distil many characteristics, finally realizes the assortment recognition, and establishes the database of main kind of trees in the northeast area, preserves the texture image and many kind of characteristic quantities, using in the panel texture characteristic video searches.But for the existence of massive disturbances noise in the texture image, the recognition rate not to be high, the operating speed slow. In view of the above question, this article starts with the texture image pretreatment, the characteristic distilling, matching and recognition, through the massive experiments, found the relative quite suitable method, finally realizes the panel assortment recognition.