Background Modeling (4) -- texture-based approach

In the previous two blog posts, we introduced the background Modeling Method Based on pixel values: the method based on pixel values (2) and the method based on pixel values (1)

Next we will introduce the texture-based feature method, which is well-known for its use of the string-based Image Processing (SLB) and the string-based Image Processing (siltp) features. This article is a texture-based method for modeling the background and detecting moving objects. This article was published on tpami in. It was also made by people from the University of ORU, they have already used the most frequently used BPS, and it has been applied in various fields of computer vision. First, perform the following calculation:

Formula:

After the feature is expressed, the background model is created, which is K. The updated model formula is as follows:

The comparison of the similarity between two histograms uses the histogram intersection (histogram intersaction). The experiment in this article is well done, but the comparison test is rarely compared with the Gaussian mixture model (GMM.
Another better method for texture-based background modeling is cvpr2010. Question: modeling pixel process with scale invariant local patterns for background subtraction in complex scenes this article proposes a new texture representation method, Scale Invariant local ternary pattern (siltp) the second is to propose a pattern Kernel Density Estimation Method (pattern Kernel Density Estimation) when the Background Modeling is updated, three methods are used to test the Nine-segment video.