Abstract
It proposes the technique of the augmented template matching for positioning the object that has two-dimensional standard shape under the outdoor environment. First of all, a whole template and some partial templates of the object are prepared. Moreover, a lot of images except the object and the object are prepared for the training data, and the linear discriminant function is decided by Fisher's linear discriminant method. Some similarities of an unknown image are calculated with the whole and partial templates. Next, similarities are substituted for the linear discriminant function, and the discriminant score is calculated. Finally, the discriminant score is judged whether it is the object by the threshold. The effectiveness of the proposed technique is shown by the corner fittings detection experiment of the container under the outdoor environment.
