Field
Buffers
Spatial
Interpolation
Motion
Detection
Temporal
Interpolation
Film Mode
Detection
Selector
Mixture
Original Line
New Line
Input
Data
Internal Modules
179
SPRUHI7A – December 2012 – Revised June 2016
Copyright © 2012–2016, Texas Instruments Incorporated
High-Definition Video Processing Subsystem (HDVPSS)
1.2.9 De-Interlacer (DEI) Module
The de-interlacer (DEI) module is used to generate progressive data output from interlaced input format
(also known as, deinterlacing).
1.2.9.1
Features
•
Motion-adaptive deinterlacing
–
Supports 4-field Motion detection
–
Motion sensitivity adaptive to the frequency of luma texture
•
Edge-Directed Interpolation (EDI)
–
Edge detection using luma pixels in a 2x7 window
–
Half-pixel resolution edge vectors
–
Edge-directed luma and chroma interpolation
–
Soft-switch between edge directed interpolation and vertical interpolation depending on the
confidence factor
•
Film Mode Detection (FMD)
–
3-2 pull down detection
–
2-2 pull down detection
–
Hysteresis controls how fast FMD can enter/exit film mode (software function)
•
Bad Edit Detection (BED)
•
The module can pass the inputs data directly to the outputs in a bypass configuration. No internal
processing is performed.
1.2.9.2
Functional Description
The following figure illustrates the block-diagram of motion-adaptive Deinterlacer. The general concept
behind motion adaptive deinterlacing is that spatial filtering works very well for images with motion, while
temporal filtering works very well for static images. So, the intuitive way is to combine them together.
Motion detection is used to switch or fade between the use of spatial deinterlacing and temporal
deinterlacing, as shown in the following formula.
ŷ
(j, i, n) =
α
y
spat
(j, i, n) + (1 -
α
)y
temp
(j, i, n)
where y
spat
(j, i, n) is the spatial interpolation output, y
temp
(j, i, n) is temporal interpolation output,
α
is the
motion detection output ranging from 0 to 1,
ŷ
(j, i, n) is the final output from deinterlacer, and j, i, n are the
vertical, horizontal, and temporal indexes, respectively. From the previous formula, the final output is
controlled by the motion detector output,
α
. The higher the motion, the higher value of
α
, and the output
favors spatial interpolation. If the motion is absent or very low, the temporal interpolation has higher
weight.
Figure 1-130. Block Diagram of Motion-Adaptive Deinterlacer