Quantizer determination, computer-readable medium and apparatus that implements at least two quantizers

Abstract

A method for determining a second quantizer for quantizing digital images, wherein the second quantizer is determined for a specified number of levels, which is at least two. For the determination, a first quantizer with a lower number of levels than the specified one is taken into consideration. Furthermore, a method for coding an image comprising a plurality of pixels, a computer-readable medium, an apparatus, which implements at least two quantizers as a digital circuit and a digital camera with such an apparatus is disclosed.

Claims

1. A method for determining a second quantizer for compressing digital image data, wherein the second quantizer is determined for a specified number of levels, which is at least two, and wherein, for the determination of the second quantizer, a first quantizer with a lower number of levels than the second quantizer is taken into consideration, wherein the method comprises determining a preliminary version of the second quantizer with the specified number of levels, and replacing those values of the preliminary version, which are closest to quantizer values of the first quantizer, with the quantizer values of the first quantizer.

2. The method in accordance with claim 1, wherein the consideration entails assuming at least one or all of the quantizer values assumed by the first quantizer as quantizer values of the second quantizer to be determined.

3. The method according to claim 1, wherein the preliminary version of the second quantizer is based on a probability density function.

4. The method according to claim 3, which includes a determination of the probability density function by averaging reference density functions belonging to at least one reference image.

5. The method according to claim 4, wherein at least one of the reference density functions approximates a relative frequency of occurrence of pixel differences, which occur in the respective reference image as pixel value differences of adjacent pixels.

6. The method according to claim 4, wherein at least one of the reference density functions approximates a relative frequency of occurrence of pixel differences, which occur in at least two related reference images as pixel value differences at a respective pixel position.

7. The method according to claim 4, wherein the reference density functions respectively belong to a family, members of which differ due to a respectively related scale parameter.

8. The method according to claim 4, wherein the probability density function is calculated at points as a weighted, arithmetical or geometric means of the reference density functions.

9. The method according to claim 7, wherein the probability density function is calculated at points as a weighted, arithmetical or geometric means of the reference density functions, wherein a weight belonging to a reference density function with its related scale parameter describes a relative frequency of occurrence respectively, by means of which the respective scale parameter occurs under the density functions belonging to the reference images.

10. The method according to claim 1, further comprising the step of implementing the determined quantizer as an electronic circuit.

11. A computer-readable medium having stored theron instructions, causing a computer carrying out the instructions to perform a method for determining a second quantizer for compressing digital image data, wherein the second quantizer is determined for a specified number of levels, which is at least two, and wherein, for the determination of the second quantizer, a first quantizer with a lower number of levels than the second quantizer is taken into consideration, wherein the method comprises determining a preliminary version of the second quantizer with the specified number of levels, and replacing those values of the preliminary version, which are closest to quantizer values of the first quantizer, with the quantizer values of the first quantizer.

12. An apparatus, which implements at least two quantizers for compressing digital image data as a digital circuit, wherein the at least two quantizers comprise a first quantizer with a first number of levels and a second quantizer with a second number of levels, wherein the second number of levels is greater than the first number of levels, wherein at least one or all of the first quantizer values assumed by a first quantizer occur as quantizer values of the second quantizer.

13. Apparatus according to claim 12, wherein the apparatus is a digital camera comprising an apparatus according to claim 12.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

(1) Preferred exemplary embodiments of the invention will be described in detail in the following based on the drawings. It is to be understood that individual elements and components can also be combined in a different manner than what is shown. Reference numbers for elements corresponding to each other are used across all figures and, if applicable, not newly described for each figure.

(2) On a schematic level, the figures show:

(3) FIG. 1a: an approach for determining a quantizer in accordance with a first exemplary embodiment of the present invention;

(4) FIG. 1b: an approach for determining a quantizer in accordance with a second exemplary embodiment of the present invention

(5) FIGS. 2a, b: an exemplary reference density and a function resulting from this in the case of the modulo operation; and

(6) FIG. 3: a digital, according to the invention in accordance with an exemplary embodiment with an exemplary image scene.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

(7) In FIGS. 1a and 1b, it is schematically illustrated how a quantizer Q2 can be determined according to the invention in accordance with a first exemplary embodiment of the present invention.

(8) Thereby, FIG. 1a shows three assignment rules, which assign respective quantizer values to elements of a (common) definition range respectively. The definition range can, for example, be given by the set of possible (meaning achievable) pixel differences between adjacent pixels of an image or between pixels of the same position in two different images; in the case of a pixel space with possible pixel values p{0, . . . , 255}, for example, the elements of the set {255, . . . , 255} are possible as differences. By means of a modulo operation (in the present example, mod 256), the definition range can be limited to the non-negative range {0, . . . , 255}, the elements of which are referred to here as modified differential values. The modulo operation can also be implicitly executed by showing the differential value in a binary manner with a specified number bits.

(9) In the example shown in FIG. 1a, a quantizer Q2 is determined with a specified number of levels 3. Thereby, in the present, initially, a preliminary version Qv of a quantizer is determined (e.g. by means of a Lloyd-Max method), the number of levels of which coincides with the specified number of levels 3.

(10) As can be recognized in FIG. 1a, the preliminary version assigns one of the values v1, v2, v3 to the modified differential values d; thereby, it has level changes (points of discontinuity) in the points b0 and b1 (of the definition range).

(11) In accordance with the illustrated exemplary embodiment, the quantizer Q2 from the preliminary version Qv is now determined taking the other quantizer Q1 with the smaller number of levels 2 under consideration (and point of discontinuity at point a0).

(12) In addition, those values of the preliminary version Qv, which are closest to a respective quantizer value of the other quantizer Q1, are then replaced with these values of the other quantizer. As can be recognized in FIGS. 1a and 1b, in the example shown, the value v1 of the preliminary version lies the closest to the quantizer value w1 of the other quantizer Q1 from the three values, of the preliminary version and the value v3 of the preliminary version lies closest to the quantizer value w2 of the other quantizer Q1.

(13) Now, the values v1 and v3 in the preliminary version Qv are replaced by the values w1 and w2, the value v2, which has a greater distance than v1 and v2, is, in contrast, maintained. Thereby, the quantizer Q2 results from this, the quantizer values w1 and w2 of which are balanced with the quantizer values of the other quantizer Q1.

(14) In FIG. 1b, it is schematically shown how a quantizer {circumflex over (Q)}2 can be determined according to the invention in accordance with an alternative embodiment: The quantizer determined in accordance with the approach shown in FIG. 1a is considered an intermediate version Qz here.

(15) From this, now, subject to maintaining the balanced quantizer values w1 and w2, the quantizer {circumflex over (Q)}2 to be determined is determined for the specified number of levels, for example, by applying the Lloyd-Max method. Thereby, in the example shown, level changes result at the points of discontinuity C0 and C1, wherein, in the present, b0<a0<c0<c1<b1. applies.

(16) The preliminary version can (in both design variants) preferably be based on a probability density function. This can result from a (e.g., weighted arithmetical) averaging of other density functions, which preferably belong to at least one reference image respectively. Such reference density functions can, for example, respectively approximate a function, which assigns the relative frequency of occurrence to the conceivable differences of two pixel values, by means of which the differences occur in the related reference image respectively.

(17) In FIG. 2a, such a reference density function f is shown. In the present example, it belongs to the family of Cauchy distributions, wherein has the scale parameter s=20. Thereby, in FIG. 2a, the set with the pixel values {0, . . . , 255} is assumed as a pixel space so thatas is mentioned above, the elements of the set {255, . . . , 255} result as possible differential values.

(18) Using the graphs of the function f, FIG. 2b shows the effect of the modulo operator mod 256: {255, . . . , 255}->{0, . . . , 255}. mod 256 (d):=d mod 256 on the incidences (of the modified differences).

(19) Thereby, a quantizer with correspondingly reduced definition range {O, . . . , 255} can be determined and used, the points of which (in contrast to the original possible differences) can each be presented with only eight bits.

(20) FIG. 3 shows a schematic illustration of a construction of a digital camera 10 according to the invention with an objective 11 as an example. Thereby, an image scene 100 is mapped over the objective 11 onto an image sensor 12, which comprises a regular arrangement of light-sensitive elements; the image scene is mapped into pixel values by means of this.

(21) The image sensor 12 transmits the pixel values as electrical data to a computing unit 13 within the camera 10, which includes, for example, a processor, a digital signal processor (DSP) or a FPGA. The computing unit 13 comprises an apparatus according to the invention, thereby implementing, in particular, at least to quantizers as a digital circuit, wherein the at least two quantizers comprise a first quantizer with a first number of levels and a second quantizer with a second greater number of levels and wherein at least one or all of the quantizer values assumed by the first quantizer occur as quantizer values of the second quantizer.

(22) The electronic data are converted into a form that can be used by the user by means of at least one of the quantizers (wherein they can, for example, be coded) and then, the data are transmitted via an interface 14 as an electronic signal 15, for example, to a receiver (not shown).

(23) A method for determining a quantizer Q2, {circumflex over ()}; Q.sub.2 for quantizing digital images, wherein the quantizer Q2, {circumflex over ()}; Q.sub.2 determined for a specified number of levels, which is at least two, is disclosed. For the determination, another quantizer Q1 with a lower number of levels than the specified one is taken into consideration.

(24) Furthermore, a method for coding an image consisting of a plurality of pixels, a computer-readable medium, an apparatus, which implements at least two quantizers Q1, Q2, {circumflex over ()}; Q.sub.2 as a digital circuit and a digital camera 10 with such an apparatus is disclosed.

(25) While at least one exemplary embodiment of the present invention(s) is disclosed herein, it should be understood that modifications, substitutions and alternatives may be apparent to one of ordinary skill in the art and can be made without departing from the scope of this disclosure. This disclosure is intended to cover any adaptations or variations of the exemplary embodiment(s). In addition, in this disclosure, the terms comprise or comprising do not exclude other elements or steps, the terms a or one do not exclude a plural number, and the term or means either or both. Furthermore, characteristics or steps which have been described may also be used in combination with other characteristics or steps and in any order unless the disclosure or context suggests otherwise. This disclosure hereby incorporates by reference the complete disclosure of any patent or application from which it claims benefit or priority.

REFERENCE LIST

(26) 10 digital camera 11 objective 12 image sensor 13 computing unit 14 interface 15 signal 100 image scene d modified differential value Q.sub.1 other (first) quantizer Q.sub.1, {circumflex over ()}; Q.sub.2 (second) quantizer to be determined Q.sub.v preliminary version Q.sub.z intermediate version a.sub.0 point of discontinuity of the other (first) quantizer Q.sub.1 b.sub.0, b.sub.1 points of discontinuity of the preliminary version c.sub.0, c.sub.1 points of discontinuity of the quantizer {circumflex over ()}; Q.sub.2 to be determined v.sub.1, v.sub.2, v.sub.3 (quantizer) values of the preliminary version w.sub.1, w.sub.2, w.sub.3 quantizer values of the other quantizer and of the quantizer to be determined