Common Target Detection Methods

Target Detection: either in each frame or when the target appears. Either the current frame information or the continuous frame information is detected. The most common method for the latter is the inter-frame difference. There are four common target detection methods:

Point detector: Moravec's operator, Harris operator, sift operator, affine invariant point Detector

Segmentation: meanshift, graph-cut, Active Contour

Background Modeling: Gaussian mixture model, eigenbackground, wall flower, dynamic texture background

Supervised classifiers: Support Vector Machine, neural network, adaptive boosting

Point Detection: Find the feature points in the image. A feature point is a point with representative texture features at its respective locations.

The most meaningful attribute of a feature point is its immutability of illumination and camera angle conversion.

Common operators include the Moravec's operator, the Harris operator, the sift operator, and the klt operator. Among them, the image deformation of the sift operator is the best.

This article from the csdn blog, reproduced please indicate the source: http://blog.csdn.net/lynphoenix/archive/2011/02/08/6174410.aspx