based on the feature detection method based on invariant technology, a scale space-based image local feature description algorithm -SIFT operator is proposed, which is stable for image scaling, rotation, affine transformation and illumination change. The generation of SIFT feature point vector consists of the following four steps: 1. Detecting extreme points in scale space; 2, removing extreme points with low contrast and unstable edge extreme points to obtain feature points; 3, calculating the direction parameters of the feature points; 4. Generate SIFT feature point vectors, the vector dimension is generally 128 dimensions. The SIFT feature point vector extracted by SIFT algorithm has the following advantages: 1. SIFT feature is a local feature of the image, which is invariant to rotation, scale scaling and brightness change, and also stable to a certain extent to visual angle change, affine transformation and noise; 2. It has good uniqueness and rich information, and is suitable for fast and accurate matching in massive feature databases; 3. Multivariate, even a few objects can produce a large number of SIFT feature vectors. The existing SIFT algorithm has some defects, and the detection efficiency and accuracy of images are poor.
technical realization idea
in view of this, the technical embodiment of this patent aims to provide an image feature detection method and device based on improved SIFT to solve the above technical problems. On the one hand, the technical embodiment of this patent provides a protection point based on improved S...
1. An image feature detection method based on improved SIFT, which is characterized by comprising: acquiring an image to be detected and a corresponding standard image; Performing image matching on the image to be detected and the standard image by using a scale-invariant feature transformation SIFT algorithm to obtain a plurality of pairs of matching points; Calculating the neighborhood diameter ratio and the direction angle difference between the matching points; According to the neighborhood diameter ratio and the direction angle difference, the matching points are eliminated to obtain correct matching points, so as to obtain features in the image to be detected.