Interpolate luma component
SSE2
| FwStatus | fwiInterpolateLuma_H264_8u_C1R | ( | const Fw8u * pSrc, Fw32s srcStep, Fw8u * pDst, Fw32s dstStep, Fw32s dx, Fw32s dy, FwiSize roiSize ); |
| dstStep | Step of the destination array. | |
| dx | Fractional part of the x-coordinate in a source image. | |
| dy | Fractional part of the y-coordinate in a source image. | |
| pDst | Pointer to a location in a destination buffer, a location in a destination array or to a destination result. | |
| pSrc | Pointer to a block of DCT coefficients. | |
| roiSize | Specifies the height and width of an ROI. | |
| srcStep | Source buffer step size (width of the buffer in bytes). |
This function increases the accuracy of inter-frame motion prediction by stepping through a defined region of a source reference frame and performing quarter-pixel luma-component interpolation. The interpolation method is convolution with a 6X6 kernel, in accordance with JVTG050 8.4.2.2.1. Results are written to a defined ROI in a destination buffer.
The function uses only the fractional portion of the complex motion vector to perform interpolation. The integer portion of the vector is used to define the source location pSrc.
To assure that all samples used for interpolation are within the boundaries of the source frame, either enlarge the source frame by using boundary samples or use the InterpolateLumaTop_H264 or InterpolateLumaBottom_H264 functions.