No self-introduction
Loading
Profile
Follow
codes (3)
Static gesture recognition
no vote
Application background In the complex background, the accurate extraction of gesture in YCbCr color space: the practice in the collection of a large number of documents and methods on the extraction of the gesture, a comprehensive comparison of various experimental simulation. Finally, in the YCbCr color space, the practice shows that: in this color space, the gesture image extraction is more complete, can meet the needs of the gesture recognition of gesture recognition. Key Technology Gesture recognition technology and its simulation: Hidden Markov model (HMM), artificial neural network, template matching method, support vector machine (SVM) is the use of artificial intelligence more recognition method. In this paper, combining with the practical application and the advantages and disadvantages of each method, selection of support vector machine for gesture recognition. Support vector machine is the first to carry out the nonlinear transformatio
25252 0 0
1
Ellipse skin color model, can run pro test
no vote
Application background Just started to try to use the skin color information to carry out the CV tracking or detection, but always found that the color information in addition to real-time results are better, the other noise, the robustness is not satisfactory. But as tornadomeet said: in fact, the ultimate realization of computer vision is a long-term process, is a classic problem in the field of AI, so the AI is a complete breakthrough, any useful information for CV is worth going deep research, unless there is an algorithm to work in all cases. Key Technology It is interesting because we use a pair of images to store the ellipse, rather than directly using the mathematical equation of the ellipse. The image is the two value image, that is, the ellipse region is white, the other is black. So when it is necessary to determine the other pixels, the pixel point is converted into Cr, Cb two coordinates, and then in the upper ellipse to find the coo
25252 0 0
1
Moving target tracking code based on particle filt
4.0
This code for particle filter movement target tracking code, personally test, absolutely good use, welcome to download. Video for the sport human body video, please run directly MAIN file debugging, parameters can adjust themselves. Forget to adopt
25252 11 0
1
No more~
Follows0
Fans0