In this section, we describe a DP algorithm implementing the recursive EDDC computation given by Equations 1 to 9, which is the basis for improvements introduced later in this paper.
Another known form for ε(p x,p y ) computation is given by Equation 12, known as the normalized cross-correlation.
where Λ = (λ transmit (1) λ transmit (2) … λ transmit (n)), and F i ′ is calculated by iterative computation using Equation (5).
By further computation of equation (4) at different time points we can obtain the following vector form equation by data point interpolation.
Therefore, we can simplify the computation in equation (8) by grouping the series of tables in Figure 12 into four types of case-control pairs, as shown in Figure 13: A-pairs (a i = 1 and b i = 1); B-pairs (a i = 1 and d i = 1); C-pairs (b i = 1 and c i = 1); and D-pairs (c i = 1 and d i = 1).
As is shown in Section 5.3, this is a significant amount of extra computation per frame compared with the ODFSD given by Equation 6.
By direct computation, the operator equation has only zero solution; then by Fredholm theorem, for any, the operator equation (3.8) has a unique solution in.
The redundancy can be removed by using the following geometric computation algorithm: solve the vertex enumeration problem for the convex polytope defined by Equation (10) and then using the obtained vertexes and rays solve the facet enumeration problem.
With a reduced feature set formed by Equation (7), we are able to speed up the computation for building a classification model, as well as avoid potential model over-fitting.
One step performs the right-sided computation of equation (8).
