Texas Instruments TMS320DM648 manual Input Filter Modes, Chrominance Re-sampling Operation

www.ti.com

Video Input Filtering

3.5.1 Input Filter Modes

The input filter has four modes of operation: no-filtering, ½ scaling, chrominance re-sampling, and ½ scaling with chrominance re-sampling. Filter operation is determined by the CMODE, SCALE, and RESMPL bits of VCxCTL.

Table 3-10shows the input filter mode selection. When 8-bit BT.656 or Y/C capture operation is selected (CMODE = x00), scaling is selected by setting the SCALE bit and chrominance re-sampling is selected by setting the RESMPL bit. If 8-bit BT.656 or Y/C capture is not selected (CMODE ≠ x00), filtering is disabled.

Table 3-10. Input Filter Mode Selection

	VCxCTL Bit
CMODE	RESMPL	SCALE	Filter Operation
x00	0	0	No filtering
x00	0	1	½ scaling
x00	1	0	Chrominance re-sampling (full scale)
x00	1	1	½ scaling with chrominance re-sampling
x01	x	x	No filtering
x10	x	x	No filtering
x11	x	x	No filtering

3.5.2 Chrominance Re-sampling Operation

Chrominance re-sampling computes chrominance values at sample points midway between the input luminance samples based on the input co-sited chrominance samples. This filter performs the horizontal portion of a conversion between YCbCr 4:2:2 format and YCbCr 4:2:0 format. The vertical portion of the conversion must be performed in software.

The chrominance re-sampling filters calculate the implied value of Cb and Cr in between luminance sample points based upon nearby co-sited Cb and Cr samples. The resulting values are clamped to between 01h and FEh and sent to the Cb and Cr capture buffers. Chrominance re-sampling is shown in Figure 3-8.

Figure 3-8. Chrominance Re-sampling

a b c d e

YCbCr 4:2:2 co-sited input samples

chroma-resampled capture results

Luma (Y)	Chroma (Cb/Cr)
- sample	samples

f g h i j k l

Cb’ef = (-3Cbc+ 101Cbe+ 33Cbg-3Cbi) / 128 Cr’ef = (-3Crc+ 101Cre+ 33Crg- 3Cri) / 128

3.5.3 Scaling Operation

The 1/2 scaling mode is used to reduce the horizontal resolution of captured luminance and chrominance data by a factor of two. For applications that require only CIF or lower resolutions, this reduces the video capture buffer memory requirements (and the bandwidth needed to write the buffer) by a factor of two.

Vertical scaling must be performed in software. (The bandwidth to load in the buffer is again reduced by 50% over the non-horizontal scaled case.)

58 Video Capture Port SPRUEM1 –May 2007

Submit Documentation Feedback