Abstract
It is difficult to recognize complex and changeable human activities in smart home, and the K-Dimensional Tree (K-D Tree) algorithm is utilized to better deal with the difficulty. The depth and computational complexity of the K-D Tree algorithm are affected by the quantity of samples. When the quantity of samples is large, the retrieval time will increase and the accuracy will decrease, so an improved algorithm- 5K-Dimensional Tree (5-K-D Tree) algorithm is proposed. Both algorithms will be applied to the field of activity recognition. The minimum Redundancy Maximum Relevance (mRMR) algorithm is also utilized in the feature selection. The results indicate that compared with the classifier based on K-D Tree, the classifier based on 5-K-D Tree increases the overall recognition rate and the single recognition rate, and also reduces the dimension of the optimal feature subset. This paper provides a new method for human activity recognition.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have