| 摘要 |
<p>PROBLEM TO BE SOLVED: To effectively reflect luminance distribution characteristics of a local area centering a FAST feature point on feature vectors, and achieve reduced calculation costs for the feature vectors to increase the speed of processing, so as to ensure sufficient accuracy and stability of matching of the feature vectors.SOLUTION: Assuming that a pixel on a small circumference that is closest in distance to a half line extending from a start point to an end point of a luminance orientation vector Ori is P0, a pixel on an extension line in a direction of the pixel P0 and on a large circumference is Q0, pixels on the small circumference counter-clockwise from the pixel P0 are pixels P1 to P7, respectively, and pixels on the small circumference counter-clockwise from the pixel Q0 are Q1 to 16, respectively, a feature vector Vf of a local area image Gx is determined by using the formula Vf=α(F00,F01,..,Ff7), where a vector component Fij is Fij=I(Qj)-I(Pi); I(x) is luminance of a pixel x; andαis a coefficient for normalizing the norm of the feature vector Vf.</p> |
