SPRUIL1D May 2019 – December 2024 DRA829J , DRA829J-Q1 , DRA829V , DRA829V-Q1 , TDA4VM , TDA4VM-Q1
Figure 6-129 illustrates the interpolation principle. The assumptions are that the input pixels are evenly spaced. The distance between each input pixel is assumed to be 1. The magnification ratio is given by 64/N and p/64 is the initial phase of the output data. The output pixels are evenly spaced as well. The distance between each output pixel is given by N/64 (in the example below, N > 64). The position of the n-th output pixel is given by (n x N + p) / 64.
Figure 6-130 illustrates the interpolation principle at the n-th output pixel (on) at position xn. Interpolation method with m and d parameters, are described as:
m = floor((n x N + p) / 64)
d = ((n x N + p) / 64) - m
Figure 6-130 illustrates the concept when the filters are configured in 4-tap mode for non-integer rescaling (the first coefficient corresponding to Om-2 is set to zero).
Figure 6-131 illustrates another explanation of the concept of a Polyphase filter using a simple 2-tap linear filter in a design which supports 5 phases.
For each one of the 5 phases, the weights (or filter coefficients) are different and after the 5 phases the weights start repeating, see Figure 6-131.
Figure 6-132 depicts the architecture for the independent polyphase filter unit.
Both the horizontal and vertical filter described earlier are an instance of the Polyphase filter unit, though only the vertical filter requires line memories (implemented in SL2 memory space). In all, there will be 20 instances of the Polyphase filter unit, 2 filters per channel and 10 total channels in the VPAC_MSC module.