SAA4945H データシートの表示（PDF） - Philips Electronics

部品番号

コンポーネント説明

メーカー

SAA4945H

LIne MEmory noise Reduction IC (LIMERIC)

Philips Electronics 

Philips Semiconductors

LIne MEmory noise Reduction IC

(LIMERIC)

Preliminary specification

SAA4945H

FUNCTIONAL DESCRIPTION

The digital LIMERIC is an effective low noise reduction IC

for luminance and colour difference signals. Noise filtering

is automatically adapted to the global noise level which is

measured within the picture content. The two dimensional

non-linear noise reduction (one for luminance, one for

chrominance) uses only line memory to function.

Furthermore, up to 10 different preferences can be set by

the user.

As shown in Fig.1, the main components of the device are

the noise reduction filter with the line memories (RAM) and

the noise estimator. Other components shown are the

reformatter, formatter, controller and a SNERT bus

transceiver.

Noise reduction filter

Both luminance and chrominance signals are filtered with

vertical recursion. This is produced as the filter receives

both filtered samples from the previous line, and unfiltered

samples from the current line. A new replacement value is

calculated for each sample read from the line memory.

This in turn, is the filtered response value for the reference

input pixel. The reference pixel is then placed at the centre

of the delay-line into which the current (unfiltered) video

line is shifted. Tables 1 to 6 show this as an ‘O’.

Both luminance and colour difference signals are filtered

using the so-called Discriminating Averaging Filter (DAF),

in which filter coefficients are related to the Absolute

Difference (AD) between samples. The filter uses samples

from both present and previous line (using the line delay)

and the result of the filter is stored back in the line memory.

In this way a vertical recursive structure is realized.

The filter coefficients are set depending on the noise

measured by the noise estimator or the NTHR (SNERT

register F9).

CHROMINANCE FILTER

The basic signal processing for either U or V is via the

same filter. It is used to process both V and U using a

multiplexed operation.

The taps structure of the chrominance filter is as shown in

Table 1.

Table 1 Chrominance processing

XXXXX

oOo

← 5 adjacent R − Y samples from the

filtered line

← 3 adjacent R − Y samples from the

incoming line

LUMINANCE FILTER

The taps structure of the luminance filter is as shown in

Table 2.

Table 2 Luminance processing

X....X...X...X....X

o.O.o

← 5 Y samples from the

filtered line (distance 4 / 5

pixel)

← 3 Y samples from the

incoming line (distance 2

pixels)

A ‘weave’ function is used to reduce any smearing effect

that could occur at edges. As shown in Tables 3 to 6, the

‘weave’ calculates over 4 consecutive lines. The relative

position of the actual pixel changes one position every line.

Table 3 For line 2n

X....X...X...X....X

o.O.o

Table 4 For line 2n + 1

X....X...X...X....X

..o.O.o

Table 5 For line 2n + 2

X....X...X...X....X

o.O.o

Table 6 For line 2n + 3

X....X...X...X....X

o.O.o..

Table 7 Weave configuration

Depending on even and odd fields the ‘weave’ has the

following configuration:

ODD FIELDS

EVEN FIELDS

X

X

X

X

X

X

X

X

X

X

X

X

X

X

1997 Jun 10

6

Share Link: