DATA SHEET
VSP 94x2A
2.3. Input Processing
UV decimation filter
5
0
3
HSYNC
5
NALPFIP
(not active
lines input)
10
15
20
25
30
35
40
Complete picture area
ALPFIP
(Active lines
input)
Active picture
0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
5
Frequency (MHz)
Fig. 2–22: UV-decimation Filter Characteristic for
Standard Operation (Decimation=1.5)
NAPPLIP
(not active
pixel per
APPLIP
(active
pixel per
line input)
line input)
2.3.2. Noise Reduction
Fig. 2–20: Image Format before Memory
The Fig. 2–23 shows a block diagram of the temporal
noise reduction. The structure of the temporal motion
adaptive noise reduction is the same for luminance as
for chrominance signal. Noise reduction is enabled by
NRON.
2.3.1. Horizontal Prescaler
The main application of the horizontal prescaler is the
conversion of the number of pixels coming form the
40.5/20.25 MHz pixel clock domain down to the num-
ber of pixels stored in the memory (factor 2/3). Gener-
ally the number of incoming pixels can be decimated
by a factor between 1 and 64 in a granularity of 2 out-
put pixels. The horizontal scaler reduces the number of
incoming pixels by subsampling. To prevent the intro-
duction of alias distortion low pass filters are used for
luminance and chrominance processing (Fig. 2–22). In
case of ITU656 input, the lowpass filter must be dis-
abled by HAAPRESC.
The output of the motion detector is weighted using the
parameters TNRCLC and TNRCLY. The look-up table
input value range is separated into 8 segments. It is
possible to freely program different behavior of the
noise reduction by using predefined curve characteris-
tic for each segment. The curve characteristics can be
programmed by the parameters TNRSxY for lumi-
nance and TNRSxC for chrominance. The curve-start
is defined by TNRSSY (TNRSSC) at the end of the last
segment (Figure 2–24). The overall curve is now con-
structed by connecting the end of segment 6 to the
beginning of segment 7 and so on. Negative values of
Ky (Kc) are not possible and clipped to zero.
The horizontal prescaler consists of two main subsam-
pling stages. The first stage is a scaler for rational dec-
imation factors in a range of 1 to 2, controlled by
HSCPRESC. The second stage decimates in integer
steps (1,2,3,4...32), controlled by HDCPRESC.
For chrominance, the result of the luminance motion
detector or a separate chorminance motion detector
can be used (TNRSEL).
Y-decimation filter
5
0
5
10
15
20
25
30
35
40
0
1.25
2.5
3.75
5
6.25
7.5
8.75
10
Frequency [MHz]
Fig. 2–21: Y-decimation Filter Characteristics for
Standard Operation (Decimation=1.5)
Micronas
Aug. 16, 2004; 6251-552-1DS
17
MICRONAS [ MICRONAS ]
