GB2314477A

GB2314477A - Image magnification processing system employing non-linear interpolation

Info

Publication number: GB2314477A
Application number: GB9612772A
Authority: GB
Inventors: Simon David Jones
Original assignee: Quantel Ltd
Current assignee: Quantel Ltd
Priority date: 1996-06-19
Filing date: 1996-06-19
Publication date: 1997-12-24
Also published as: JPH1063828A; GB9612772D0; EP0814429A2; EP0814429A3

Description

2314477 IMPROVEMENTS.REL,1TIWG-TO IMAGE PROCESSING This invention relates

to image magnification means and more particularly to such means for magnifying an image formed by an array of pixels each comprising digital data by providing, by interpolation between adjacent pixels thereof, pixels of a magnified form of said image.

In our United Kingdom Patent No. 2 089 625 B and corresponding United States Patent No. 4 514 818, the teachings of which are herein incorporated by reference, a video graphics system is described which includes a touch tablet and stylus combination for generating position signals designating the position of the stylus relative to the touch tablet. The user or operative paints or draws (hereinafter referred to simply as "paints") by selecting a colour and a, so-called brush distribution from a range of pre-defined colours and distribution functions. Movement of the stylus relative to the touch tablet causes the generation of a series of position signals.

In equipment according to our British Patent No. 2 089 625, as currently manufactured and sold by the Assignee of this application, additional means are provided which allow the system to be configured to perform picture composition in addition to painting. Some configurations which allow composition are described in British Patent No. 2 113 950 and in corresponding US Patent No. 4 602 286 the teachings of which are also incorporated herein by reference.

When such a system is required to effect painting on a high resolution graphic image, e.g. 10,000 x 8,000 pixels, the largest normally available brush size which is limited to 200 1 pixels diameter appears very small in relation to the overall image and the painting of large areas of the image is therefore a slow process.

One way of dealing with this problem would be to provide a brush store with much larger brush stamp sizes but that solution would involve having a brush store of greatly increased capacity which would prove expensive.

The problem is usually dealt with by having as shown in Figure 1 (a) - (f) a special mode of operating the system according to which a reduced resolution version of the image is displayed on a monitor. Suitably the displayed image of Figure 1(a) is reduced by a factor of eight in both the horizontal and vertical directions to provide an image of approximately 1250 x 1000 pixels as seen in Figure 1(b) which can be displayed on an HDTV monitor which is of comparable resolution. The user paints over this image using the brush sizes which are normally available (i.e. up to 200 pixels in diameter) but the system actually paints into a special stencil seen in Figure 1(c) through which the user can see the displayed image. This stencil has the same resolution as the displayed image. The brush stamps which make up the brush stroke thus appear much larger in relation to the displayed image than they do in relation to the high resolution image. When painting is completed, the reduced resolution displayed image which was only needed for display purposes is discarded leaving the painted stencil of Figure 1(d) which is magnified up to the size of the high resolution image as seen in Figure 1(e). The magnified painted stencil is finally keyed over the high resolution image as illustrated in Figure 1 (f).

2 Figure 2(a) illustrates the brightness (or colour) - labelled "amplitude" - distribution of a small brush stamp 1 in which the brightness (or colour) is approximately constant in a central part 3 and falls away to zero at the edges 5 and 7 of the stamp. Sampling points are indicated by crosses 9 on the brush distribution profile 11. Figure 2(b) illustrates a magnified form of the brush stamp of Figure 2(a) obtained by a sizing / efFects engine within the apparatus and of the kind described in our co- pending patent application No. 9518443.8 dated 8 September 1995 the contents of which are incorporated herein by reference. In Figure 2(b) interpolated pixels 12 are present between successive sampling points 9.

The magnification of the stencil creates an enlarged brush stamp 13, having additional sampling points therein, generated by interpolating between samples of the original stamp 1.

The usual interpolation method employed to create the further samples 12 is bi-linear 4-point interpolation in which an interpolated pixel is derived from four adjacent pixels. In the magnified stamp form of Figure 2(b) this is equivalent to a straight line approximation between successive samples as shown in Figure 3. The problem with this method of interpolation is that in the soft edge of the magnified brush stamp 13 where the brightness (or colour) falls away, there appears a series of bands, as illustrated in Figure 4 where the bands are indicated by references a, b and c and are of diminishing brightness. This banding is attributable to the step change in brightness (or colour) gradient between the interpolated line segments, i.e. at the sampling locations. This type of change is perceived 3 by the eye as an edge even though it is a second order artifact. In this context a first order artifact is one which occurs at a step change in the brightness (or colour) at a transition in an image and a second order artifact is one which occurs where a step change in the rate of change brightness (or colour) occurs at a transition in the image.

It is an object of the invention to provide image magnification means which effect improved image magnification in which second order artifacts occurring in the image are diminished or eliminated.

The present invention consists in image magnification means for magnifying an image formed by an array of pixels each comprising digital data by providing, by interpolation between adjacent pixels thereof, pixels of a magnified form of said image, comprising processing means which employ non-linear interpolation functions to calculate interpolated pixel values in said magnified image which diminish the prominence at least of second order artifacts in said magnified image.

Suitably, said non-linear interpolation functions are spline functions. Advantageously, the spline functions employed are local control spline functions. Preferably, the local control functions are B-spline functions.

The invention further consists in image magnification means for magnifying an image formed by an array of pixels each comprising digital data by providing, by interpolation between adjacent pixels of said image, pixels of a magnified form thereof, comprising a memory in the form of a look-up table stored with filter coefficients representative of 4 spline weighting functions and a calculation means adapted to employ said coefficients to calculate interpolated pixel values in the magnified image such that at least second order artifacts in the magnified image are diminished.

Suitably there is provided a magnification and interpolation control which selects the spline weighting function coefficients to be employed by said calculation means. Suitably, also said image and said magnified image are stored in respective source image and destination image stores and said magnification and interpolation control instructs respective source and destination address generators to locate source image pixels required for calculation of a corresponding destination image pixel and to calculate the address of said destination pixel.

The invention will now be described, by way of example, with difference to the accompanying drawings, in which:- Figure 1 (a) - (f) is a diagrammatic view illustrating a known procedure for applying a large brush stroke to a high definition image of 1000 x 8000 pixels, Figure 2(a) and 2(b) illustrate respectively a brush distribution function and a magnified version thereof, Figure 3 illustrates to a larger scale the linear interpolated form of the magnified brush distribution function of Figure 2(b) Figure 4 illustrates the artifacts in a brush stroke attributable to the magnified brush distribution of Figure 2(b) Figure 5 is a block diagram of an image magnifying means according to the invention.

In the drawings like parts are accorded the same references.

Figures 1 to 4 of the drawings have been discussed above.

Referring now to Figure 5, this depicts a simplified though more generalised form of the block diagram comprising Figure 1 of our copending application No. 9518443.8, and illustrates an image magnifying means 20 according to the present invention.

A source store 21 receives video image data from any suitable external source (not shown) such as a video tape, camera or disc store and a calculating mechanism 23 receives pixel 6 data from the store 21 by way of a data bus 25 on which it is to perform a magnifying effect by the creation of interpolated pixel data. The magnified data which may or may not require the presence of unmodified source pixel data as well as interpolated data is output on data bus 27 from the mechanism 23 to a destination image store 29. A source address generator 31 determines the pixels and the order thereof to be supplied from the source store 21 to the calculation mechanism and a destination address generator 33 determines the address in the destination store to which the magnified image pixel data from the calculation mechanism is to be directed.

In addition to the source pixel data received from the source store 21, the calculating mechanism also receives weighting coefficients from one of a number of spline functions which, preferably, are B-spline functions which are applied to the source pixels supplied to the calculating mechanism to provide, as hereinafter described in further detail, the output pixels supplied to the destination store 23. Although B-spline functions are those, of choice, employed in the preferred form of the invention it is pointed out that any non-linear interpolation function which has the effect of diminishing the visibility in the destination image of second order artifacts may be employed.

The spline weighting function coefficients are stored in a memory comprising a look-up table (LUT) 35 and the coefficients of the spline employed by the mechanism 23 are selected by a magnification and interpolation control 37 which also instructs the source and destination address generators 31 and 33 respectively to generate the pixels addresses 7 which are required for the magnification calculation and the pixels addresses at which the pixels of the magnified image are to be stored.

In its turn the magnification and interpolation control 37 is under the control of a top level controller 39 which instructs the control 37 as to the type, e.g. B-spline, and the degree of magnification operation to be effected.

Preferably, the magnifying means operates as a read side system. Thus the control 37 calculates for each pixel of the image in the destination store the pixel or pixels of the image in the source store which contribute to the value of that pixel of the destination store image. The source image pixels so identified are then supplied to the calculating mechanism.

The calculating mechanism suitably could be of the form described with reference to Figures 10 and 12 of our co-pending patent application No. 9518443.8 though the usage herein is for the purpose of magnification rather than filtering. In such a form, as in the patent application referred to, the calculation mechanism comprises three (or more) multipliers to each of which is supplied a pixel and a spline coefficient. The products of the multipliers are added in adder and the result is sent to an accumulator which accumulates the results of each calculation performed together by the multipliers. The number of simultaneous calculations carried by the multipliers to evaluate a particular destination image pixel depends on the number of source image pixels which require to be 8 addressed. The spline coefficients are fractional values which in the overall calculation of each pixel of the destination image total unity.

Brightness, Y, and colour difference U and V values of the destination image pixels are separately calculated in the manner described to provide the total data for each destination image pixel. The destination image can be sent to a display monitor, stored in a disc store or on a tape or printed.

Although a look-up table 35 has been referred to for storage of the spline coefficients, the latter could be calculated "on the fly" by the magnification and interpolation control and supplied directly to the multipliers. The use of B-spline weighting functions is preferred as they effect local smoothing of the brush profile curve by calculation of each interpolated pixel using a few, suitably about three successive brush stamp samples 9 and the calculation engine for achieving this conveniently is a modified form of the finite impulse response filter disclosed in our co-pending application No. 9518443.8 referred to above.

9

Claims

1. Image magnification means for magnifying an image formed by an array of pixels each comprising digital data by providing, by interpolation between adjacent pixels thereof, pixels of a magnified form of said image, comprising processing means which employ non-linear interpolation functions to calculate interpolated pixel values in said magnified image which diminish the prominence at least of second order artifacts in said magnified image.

2. Image magnification means as claimed in claim 1, wherein said nonlinear interpolation functions are spline functions.

3. Image magnification means as claimed in claim 2, wherein said spline functions are local control spline functions.

4. Image magnification means as claimed in claim 3, wherein said local control spline functions are B-spline functions.

5. Image magnification means for magnifying an image formed by an array of pixels each comprising digital data by providing, by interpolation between adjacent pixels of said image, pixels of a magnified form thereof, comprising a memory in the form of a look-up table stored with filter coefficients representative of spline weighting functions and a calculation means adapted to employ said coefficients to calculate interpolated pixel values in the magnified image such that at least second order artifacts in the magnified image are diminished.

6. Image magnification means as claimed in claim 5, wherein there is provided magnification and interpolation control means which select the spline weighting function coefficients to be employed by said calculation means.

7. Image magnification means as claimed in claim 6, wherein there are provided respective source and destination stores for storing said image and said magnified image and said magnification and interpolation control is adapted to instruct respective image and magnified image address generators, in the case of the image address generator to locate image pixels required for calculation of a corresponding magnified image pixel and in the case of the magnified image address generator to calculate the address of said corresponding magnified image pixel.

8. Image magnification means as claimed in claim 6 or claim 7, wherein there is provided a look-up table for storing said spline weighting function coefficients.

9. Image magnification means as claimed in claim 8, wherein said look-up table is adapted to store a plurality of spline weighting functions and said magnification and interpolation control servers to select the function coefficients of one of said functions for employment by the calculation means.

10. Image magnification means as claimed in any one of claims 5 to 9, wherein said spline weighting functions are B-spline, local control functions.

11 11. Image magnification means as claimed in any one of claims 5 to 10, wherein said calculation means comprises a modified finite impulse response filter.

12. Image magnification means substantially as herein before described with reference to the accompanying drawings.

12