Television – Format conversion – Line doublers type
Reexamination Certificate
1999-03-04
2002-09-17
Miller, John (Department: 2614)
Television
Format conversion
Line doublers type
C348S441000, C348S451000, C348S452000, C382S300000
Reexamination Certificate
active
06452639
ABSTRACT:
The invention relates to a method and a system for an adaptable scan raster conversion of an interlaced source signal into a progressive output signal by use of a signal interpolation algorithm (IPC algorithm).
Interlaced to progressive conversion interpolation algorithms interpolate missing lines in an interlaced scanned source field in order to create a progressive frame. Interlaced scanning is a spatio-temporal sub-sampling scheme, i.e. vertical sub-sampling in the spatial dimension. This sampling raster combines the advantages of a high temporal sampling frequency, i.e. high field rate, with the advantages of a low spatial sampling rate with a moderate bandwidth or data rate.
For progressive display media, e.g. computer monitors, LDC displays, etc., the problem is, to display an “interlaced signal” in a progressive fashion. Before this can be done, a formate conversion or scan raster conversion has to be performed, which transforms the signal from interlaced to progressive form (IPC). Since interlaced scanning is a spatio-temporal sub-sampling as mentioned, a spatio-temporal interpolation scheme has to be applied in order to achieve the best possible results.
IPC algorithms can be coarsely divided into three algorithm classes. This classification is based on the most critical and characteristic source signal, i.e. image motion. The classes are:
static, i.e. no motion information whatsoever;
motion adaptive, i.e. a motion detector with some ON/OFF selection is required;
motion compensated, i.e. it includes motion vector information.
Static algorithms are basically motion-insensitive. i.e., time-invariant filter schemes and lack a distinct motion detector or motion vector estimator. Examples are:
line repetition,
linear line interpolation, i.e. one-, two- or three-dimensional FIR filters with a fixed spatio-temporal filter aperture, and
non-linear median filters, i.e. one-, two- or three-dimensional filters.
Due to the simplicity of these algorithm classes, the filters have a relatively simple implementation, but their visual performance is potentially limited in comparison to motion adaptive or motion compensated algorithms. A good spatio temperal behaviour of an IPC algorithm, according to the state of the art knowledge, is only achievable with a motion vector estimation.
Motion adaptive algorithms feature a motion detector component, which performs soft- or hard-switching between mere spatial interpolation or a spatio/temporal interpolation, i.e. filter mask selection. The switching scheme includes thresholds and defined switching method. In other words, a motion signaling switches between spatial and spatio-temporal interpolation, i.e. the filter mask selection switching between vertical and temporal interpolation, and the motion signal is detected on a frame or field basis. A variety of motion detectors are proposed in the literature and reference is made to the literature list attached to this description.
However, a motion detector estimation is a computationally expensive task.
It is an object of the invention to provide a new and improved method and system for a scan raster conversion of an interlaced source signal into a progressive output signal with good visual results but with low computational efforts.
An implementation related problem is the availability of system resources which can be used for the IPC algorithm. In a complex image processing system, which serves multiple purposes, resource availability is often a time-variant function. This problem is approached with the invention by use of a scaleable IPC algorithm which—depending on the available resources at a certain time—yields the best possible interpolation result.
Based on this background from the above, the IPC problem is solved by the invention by an IPC algorithm wich converts an interlaced source signal into a progressive output signal with computational costs limited to a minimum by use of only static but scaleable IPC algorithm depending on the time-variant availability of system resources.
According to the invention, a method for an adaptable scan raster conversion of an interlaced source signal into a progressive output signal by use of a signal interpolation algorithm (IPC algorithm) is proposed which is scaleable in terms of various modes depending on the availability of the respective system resources and/or external constrains at a certain time.
Said system resources may comprise an alternatively scaleable IPC system with respectively associated alternative processing modules and/or module resources, respectively selectable depending on a grouped selection by an IPC mode control. Such system resources and/or external constrains may be computational power, available memory capacity, available memory bandwidth. Preferably, said interpolation algorithm in respect to each selectable mode is of a linear filter type and/or a median type and/or a combination of a linear interpolation, e.g. FIR filtering and a median filtering.
A scan raster conversion system for converting an interlaced source signal into a progressive output signal by use of a signal interpolation algorithm (IPC algorithm) comprises according to the invention
various modules for signal interpolation, and
an IPC mode control means provided with knowledge about required resources for specific classes of interpolation processes and receiving input information about available resources of an embracing system at a certain time for supplying command structures for a mode specific selection of one or a combined plurality of said interpolation modules dependent on said input information.
According to a preferred embodiment, said various modules for signal interpolation comprise at least one linear interpolator and a ranking filter interpolator.
The various signal interpolation modules are based on the one hand on linear algorithms that can be divided into source signal insensitive algorithms, such as the above mentioned linear filters, and in non linear signal adaptive algorithms, such as
median filters,
weighted median filters,
vertical and/or diagonal edge adaptive interpolators, including edge adaptive median filters, and
filters that adapt to the spatial slope of lines.
This last mentioned third class is the most elaborate interpolation scheme and offers the possibility of maintaining high spatial resolutions and thus a high visual quality of the interpolation results, i.e. of the progressive frame with minimal aliasing with a variety of motion speeds.
An alternative class of algorithms, the motion compensated algorithms, features a variety of components, i.e.:
motion vectors estimation on a frame or field basis,
motion vectors are computed for a variety of block sizes (spatial granularity) which can be as small as a single pixel, i.e. a motion vector for each source pixel,
motion vector precision can vary and can be as detailed as on a sub-pixel basis,
the interpolation scheme, which estimates the missing pixels in an interlaced source field incorporates this motion vector information to enhance a visual quality of a resulting progressive image.
A variety of motion estimators has been proposed such as
block search,
recursive block search/block matching,
phase correlation black matching,
schemes to adapt MPEG motion vectors in order to use them for IPC.
The algorithm structure or architecture according to the invention is adaptive depending on the current resource setting or availability. This algorithm in its various modes belongs to the static interpolator type in consideration of implementation restrictions such as CPU power and memory. In its advanced mode, it is the mixture of a linear filtering, i.e. linear interpolation part, and edge adaptive median filtering, implemented in three-tap and five-tap median processor part.
REFERENCES:
patent: 5483288 (1996-01-01), Hong
patent: 5493338 (1996-02-01), Hong
patent: 5943099 (1999-08-01), Kim
patent: 5953465 (1999-09-01), Saotome
patent: 5991464 (1999-11-01), Hsu et al.
patent: 6141056 (2000-09-01), Westerman
patent: 6157749 (2000-12-01), Miyake
patent: 6262773 (2001-07-01), West
Erdler Oliver
Schwendowius Joerg
Wagner Peter
Zimmermann Klaus
Desir Jean W.
Frommer William S.
Frommer & Lawrence & Haug LLP
Miller John
Simon Darren M.
LandOfFree
Raster scan conversion system for interpolating interlaced... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Raster scan conversion system for interpolating interlaced..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Raster scan conversion system for interpolating interlaced... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2907950