Based On The Research Of The Bag-of-Words Model Wood Classification

In the process of modern production,the low degree of automation manufacturers,whose most of the wood classification jobs are made by artificial,having problems on classification accuracy and efficiency.In order to improve the market common wood accurate recognition rate,we introduced' Bag-Of-Words(BOW)model',which in the field of image classification developing rapidly and having good effect on the classification filed to solve the wood classification problems.Meanwhile,pyramid space matching and the histogram cross nuclear is introduced to the Bag-Of-Words model,BOW ? pyramid space matching and the histogram cross nuclear matching traversal combination,BOW model based on histogram intersection kernel(BOW-H),BOW model based on spatial pyramid matching(BOW-P),BOW based on spatial pyramid matching and histogram intersection kernel(BOW-HP)four kinds of classification methods to classify the test.Compared with The BP neural network and the kneighbor classification method,the experimental data obtained are compared and analyzed in the same experimental environment.Northeast five common trees named Birch Radial section ? Birch Tangetial section ?Korean pine Radial section ? Korean pine Tangetial section ? Larch Radial section ? Larch Tangetial section ? Ashtree Radial section ? Ashtree Tangetial section ? Oak Radial section and Oak Tangetial section.Almost ten kind of images,each type of 100,a total of 1000 images as sample.,Using SIFT algorithm to extract image features,using K-means clustering algorithm for clustering,The Bag-Of-Words model and its improved method are very accurate.In the same experimental environment,the classification accuracy is higher than the BP neural network and k-nearest neighbor.It is proved that it is effective to introduce the Bag-Of-Words model and its improvement algorithm into single board classification...