Method and Apparatus for Video Signal Interpolation and Progressive Scan Conversion

This invention relates generally to television systems and particularly to methods and apparatus for video signal interpolation and for conversion of video signals from an interlaced form to a non-interlaced or "progressive scan" form.

In image processing, there are various methods for converting video signals from interlaced to progressive scan form. Typically, these employ some form of spatial and/or temporal interpolation to obtain "extra" lines for display from the incoming video signal. For example, there are television receivers which double the lines in one field by either repeating lines or obtaining the extra lines needed by means of vertical (spatial) interpolation. Such forms of conversion are commonly called intra-field progressive scan conversion systems since the "extra" lines for display are obtained by interpolation from lines of the same field.

Other forms of conversion are known which involve both spatial and temporal interpolation. In such systems lines from temporally associated fields are used in forming a progressive scan signal in cases where the image has no motion (a static image). A motion detector determines the presence or absence of motion and, for the case where motion exists, the extra lines for display are obtained by vertical interpolation. Progressive scan converters of this type is generally known as motion adaptive or motion "compensated" converters since the decision to use temporal or spatial interpolation is determined by the motion detector.

In the first case (intra-field conversion), the hardware complexity is relatively low but the resultant image quality is not optimum with certain image contents. In the second case, (motion adaptive conversion) the image quality is higher but the hardware complexity is also higher. A further problem with conventional motion adaptive systems is that errors in the detection of motion can produce visible artifacts when switching between field and frame processing.

The present invention is directed to meeting the need for a method and apparatus for providing progressive-interlace conversion in which good image quality is provided with relatively low hardware complexity.

In accordance with the invention, a median filter is used for carrying out for each pixel to be determined a temporal interpolation of two temporally adjacent pixels of the same spatial position. The result is corrected by means of a further median filtering of the vertically adjacent pixels. For this double median (hereafter, "DM") interpolation it is of advantage that no motion detection is required. In addition, the DM interpolation is relatively insensitive to noise.

A method for interpolating a video input signal, in accordance with the invention, comprises forming a vertically interpolated signal from the video input signal; forming a first median filtered signal by applying the vertically interpolated signal, a field delayed signal and a field advanced signal to respective inputs of a first median filter; and forming a video output signal subjected to double median filtering by applying said first median filtered signal, a line delayed signal and a line advanced signal to respective inputs of a second median filter.

In an exemplary embodiment of the invention, apparatus is provided having an input from which interlace pixels are supplied to a first field buffer and to a first median filter, which also receives the output signal of a second field buffer and the output signal of an averaging circuit, the output signal of the first field buffer being supplied to a line buffer, to the first input of the averaging circuit and to a second median filter, and the output signal of the line buffer passing to the input of the second field buffer, to a second input of the averaging circuit and to a second median filter which is also fed with the output signal of the first median filter, the values for the pixels of the progressive output signal being taken at the output of the second median filter.

The foregoing and further features of the invention are illustrated in the accompanying drawing, wherein:

FIG. 1 illustrates a vertical-temporal scanning diagram;

FIG. 2 is a pixel diagram illustrating pixels for interpolation in accordance with the invention;

FIG. 3 is a block diagram illustrating a double median filter interpolation processor according to the invention;

FIG. 4 is a block diagram illustrating application of the double median processor of FIG. 3 for luminance signal processing in a progressive scan processor;

FIG. 5 is a block diagram illustrating application of the double median processor of FIG. 3 for chrominance processing in a progressive scan processor; and

FIG. 6 is a block diagram illustrating a vertical averaging chrominance progressive scan processor suitable for use with the double median luminance progressive scan processor of FIG. 4 in a progressive scan conversion system.

FIG. 1 shows scanned pixels in the vertical-temporal (v, t) plane. The small circles mark the pixels of the interlace input signal, for example 625/50/2:1 (i.e., 625 lines per frame, 50 fields per second, two to one line interlace). The large circles mark the pixels of the progressive output signal, for example 625/50/1:1. A first field belongs to time t=0 ms and a second field belongs to time t=20 ms.

Pixels P.sub.2A and P.sub.2B in FIG. 2 represent the pixels of the progressive output signal which are to be determined in each case. The luminance and chrominance value of pixel P.sub.2B is obtained by means of two median filterings from the corresponding values of the two vertically adjacent pixels (P.sub.2, P.sub.3) and the two temporally adjacent pixels (P.sub.1, P.sub.4) of the interlace input signal. The value of pixel P.sub.2A is transferred unchanged from pixel P.sub.2. The value for P.sub.2B is determined in accordance with the following formula:

With a static image content (P.sub.1 =P.sub.4), the value for P.sub.2B is initially equal to P.sub.1 or, respectively, P.sub.4 due to the first temporal median filtering. Although the second vertical median filtering suppresses the theoretically highest vertical frequency which can be transmitted in the frame (frame lines alternately white-black-white), this image content does not occur in practice and would also lead to strong interline flickering in the case of a normal interlace display. It is much more important that vertical image signal amplitude steps are reproduced with full sharpness and that at the same time, nevertheless, no 25 Hz interline flickering occurs on horizontal edges.

With a dynamic image content, the temporal resolution is slightly reduced by the first median filtering. However, should the value for P.sub.2B initially determined deviate too much from the vertically adjacent image content, it is automatically corrected by the second median filtering (post test).

In FIG. 3, interlace pixels P.sub.4 are supplied at the input Y to a first 312-line buffer (field buffer) 31 and to a first median filter ME1, which also receives the output signal P.sub.1 of a second 312-line buffer 32 and the output signal (P.sub.2 +P.sub.3)/2 of an averaging circuit AV. The output signal P.sub.3 of the field buffer 31 is supplied to a line buffer 33, to one input of the averaging circuit AV and to a second median filter ME2. The output signal P.sub.2 of the line buffer 33 passes to the input of the second 312-line buffer 32, to the other input of the average circuit AV and to the second median filter ME2 which is also fed with the output signal of the first median filter ME1. At output O, the values P.sub.2B for the pixels of the progressive output signal can be picked up.

The values for the chrominance pixels can be obtained either by corresponding double median filtering or by vertical averaging (averaging circuit AV) in the field with the aid of a line buffer (33) with the appropriate temporal adaptation to the luminance: P.sub.2A =P.sub.2, P.sub.2B =(P.sub.2 +P.sub.3)/2.

FIG. 4 illustrates use of the double median processor of FIG. 3 in a progressive scan system. A luminance input signal Y1 is applied to the input of a double median processor 400 which outputs a video signal Y2 processed by double median filtering as previously described and a field delayed luminance signal Y3 (e.g., delayed by delays 31 and 33). The delayed Y3 and interpolated Y2 signals are time compressed by a factor of 2:1 by respective time compressors 404 and 402 and applied to a multiplex switch 406 which interleaves the time compressed lines of signals Y4 and Y5 to form the progressive scan output signal Y6 in which every other line corresponds to the delayed input signal and the in-between lines are interpolated by double median filtering.

As noted above, the chrominance signal C1 may be processed by double median filtering also and this is shown in FIG. 5 in which the chrominance signal C1 is applied to a double median filter 500 (as per FIG. 3) to generate a field delayed chrominance signal C3 and a double median filtered interpolated chrominance signal C2. These signals are time compressed in respective time compressors 504 and 502 and the speeded-up signals C4 and C5 and interleaved by a multiplex switch 506 to generate the progressive scan chrominance output signal C6. As before, every other line of this signal corresponds to the input chrominance signal (delayed by a field to align with the luminance signal) and the in-between lines correspond to the chrominance signal interpolated by double median filtering.

An advantage of using double median filter processing of both the luminance and chrominance components in a progressive scan system is that the color detail is of very high quality. However, since detail is conveyed mainly by the luminance signal one may simplify the progressive scan system by applying double median filtering to the luminance signal (as per FIG. 4) and by applying vertical averaging to the chrominance signal. A suitable chrominance signal processor using vertical averaging and having a proper temporal response to align with the double median filtered luminance signal is shown in FIG. 6. There, a chrominance signal C10 is delayed by a field delay 600 and applied to a line delay 604 and to an averager 602 which receives at its other input the line delayed signal C12. The outputs of the line delay and the averager are applied to respective time compressors 608 and 606 and the compressed signals are interleaved to form a progressive scan video output signal by means of multiplex switch 610.

The invention can be used, for example for the PAL, SECAM, NTSC, (D2)MAC, HDMAC and MUSE television standards with corresponding numbers of lines in television standards with corresponding numbers of lines in television receivers and video recorders. The said numerical values are then changed as appropriate.