SPRUIE9D May 2017 – May 2024 DRA74P , DRA75P , DRA76P , DRA77P
Figure 9-101 shows the basic interpolation method used in the RSZ module. The following assumptions are made:
The output pixels are also evenly spaced. The distance between each output pixel is given by N/256. In the example in Figure 9-101, N is greater than 256. The position of the nth output pixel is given by (n × N + p) / 256.
Figure 9-101 ISS ISP RSZ Basic Interpolation MethodAssuming the intensity of input pixels is i0, i1, i2, etc. and the resized (output) pixels are o0, o1, o2, etc., the nth output pixel (on) is determined using the nearest 4 input pixels as follows:
In the previous equation h(x) is the interpolation main function. The RSZ module supports linear and bicubic convolution interpolation functions.
Figure 9-102 shows the interpolation principle at the nth output pixel (on) at position xn. Furthermore, the m and d parameters are as follows:
Figure 9-102 ISS ISP RSZ Interpolation FilteringFor each resizer (RSZ-A or RSZ-B), and for Chroma and Luma, the interpolation method for vertical interpolation can be a 2-tap linear interpolation or a 4-tap cubic convolution (default) method. The choice is made in RZi_V_TYP[0] Y for Luma and RZi_V_TYP[1] C for Chroma, where i = A or B and is the resizer number. It is similar for horizontal interpolation from RZi_H_TYP[0] Y for Luma and RZi_H_TYP[1] C for Chroma, where i = A or B and is the resizer number.