T 2249/12 () of 8.6.2017

European Case Law Identifier: ECLI:EP:BA:2017:T224912.20170608
Date of decision: 08 June 2017
Case number: T 2249/12
Application number: 02738883.4
IPC class: H04N 7/26
H04N 7/32
H04N 7/46
H04N 7/50
Language of proceedings: EN
Distribution: D
Download and more information:
Decision text in EN (PDF, 475.490K)
Documentation of the appeal procedure can be found in the Register
Bibliographic information is available in: EN
Versions: Unpublished
Title of application: Image encoder, image decoder, image encoding method, and image decoding method
Applicant name: NTT DoCoMo, Inc.
Opponent name: -
Board: 3.5.04
Headnote: -
Relevant legal provisions:
European Patent Convention 1973 Art 84
European Patent Convention Art 123(2)
European Patent Convention 1973 Art 56
Keywords: Claims - clarity
Claims - main request, first to fifth auxiliary requests (no)
Amendments - extension beyond the content of the application as filed
Amendments - sixth auxiliary request (yes)
Inventive step - amended sixth auxiliary request (no)
Inventive step - seventh auxiliary request (yes)
Catchwords:

-

Cited decisions:
-
Citing decisions:
-

Summary of Facts and Submissions

I. The appeal is directed against the decision to refuse European patent application No. 02 738 883.4.

II. The patent application was refused on the grounds that the independent claims of all requests then on file did not comply with Article 84 EPC, and that the claimed subject-matter did not involve an inventive step within the meaning of Article 56 EPC in view of documents

D1: US 5 126 841 A and

D4: VCEG-M81: "H.26L Test Model Long-Term Number 7" 13th VCEG MEETING; 2-4 April 2001; Austin, Texas, US; (Videocoding experts group of ITU-T SG. 16), no. VCEG-M81d0, 11 May 2001, XP030003250, ISSN: 0000-0461.

III. The applicant appealed against this decision and with the statement of grounds of appeal submitted claims of a main request and of first to ninth auxiliary requests. It submitted that the claims of the main and the first to fifth auxiliary requests were identical to the requests underlying the decision under appeal.

IV. The board issued a summons to oral proceedings and indicated in a communication annexed to the summons that it considered claims 1 of the main and first to fifth auxiliary requests to be unclear. With respect to claim 1 of the sixth auxiliary request the board raised inter alia an objection under Article 123(2) EPC and indicated that the subject-matter of the independent claims of the sixth auxiliary request lacked inventive step in view of D1 combined with D4.

V. In a letter of reply dated 3 May 2017 the appellant provided arguments addressing the board's observations and submitted claims of a tenth, an eleventh and a twelfth auxiliary request.

VI. Oral proceedings before the board were held on 8 June 2017. During the oral proceedings the appellant submitted claims 1 to 17 of an Amended Sixth Auxiliary Request and claims 1 to 18 of a Seventh Auxiliary Request named "Amended Auxiliary Request 7". The appellant requested that the decision under appeal be set aside and that a European patent be granted on the basis of the claims of

- the Main Request or the First to Sixth Auxiliary Requests, all filed with the statement of grounds of appeal, or

- the Amended Sixth Auxiliary Request filed during the oral proceedings of 8 June 2017, or

- the Seventh Auxiliary Request filed during the oral proceedings of 8 June 2017.

VII. Claim 1 of the main request reads as follows:

"An image coding apparatus, characterized by comprising:

motion compensation predicting means for generating a motion-compensated prediction image by detecting movement amounts in predetermined partial image region units of an input image;

smoothing means for performing smoothing of pixels located at the boundaries of adjoining partial image regions on the basis of predetermined evaluation criteria, with respect to the prediction image obtained by the motion compensation predicting means; and

prediction residual coding means for coding the prediction residual signal obtained from the difference between the input image and the smoothed prediction image; wherein the smoothing means comprise

activity level setting means for determining the activity level between the adjoining partial image regions on the basis of a difference in coding mode indicating inter coding or intra coding between the adjoining partial image regions;

adaptive smoothing means for establishing the intensity of the smoothing on the basis of the activity level and then performing smoothing of pixels located at the boundaries between the partial image regions."

VIII. Claim 1 of the first auxiliary request contains the following additional feature appended to claim 1 of the main request:

"..., wherein the pixels targeted for smoothing are changed in accordance with the activity level."

IX. Claim 1 of the second auxiliary request corresponds to claim 1 of the main request, but with a deletion in the feature relating to the activity level setting means (shown by strikethrough below):

"... activity level setting means for determining the activity level between the adjoining partial image regions on the basis of a difference in coding mode [deleted: indicating inter coding or intra coding] between the adjoining partial image regions;".

X. Claim 1 of the third auxiliary request contains the following additional feature appended to claim 1 of the main request:

"...; and post-processing means for performing suppression processing of the results of smoothing by means of a threshold value established in accordance with the activity level wherein the differential values between the pixel values before the smoothing and the pixel values after the smoothing are controlled by means of the threshold value."

XI. Claim 1 of the fourth auxiliary request contains the following additional feature appended to claim 1 of the third auxiliary request:

"... wherein the threshold is the maximum permissible differential value and in cases where the differential value between a pixel value before the smoothing and a pixel value after the smoothing is equal to or more than the threshold value, the pixel value after the smoothing is corrected to diminish the differential value in accordance with the size of the threshold value."

XII. Claim 1 of the fifth auxiliary request corresponds to claim 1 of the second auxiliary request with the additional feature of claim 1 of the third auxiliary request appended to it.

XIII. Claim 1 of the sixth auxiliary request reads as follows:

"An image coding apparatus, characterized by comprising:

motion compensation predicting means for generating a motion-compensated prediction image by detecting movement amounts in predetermined partial image region units of an input image;

smoothing means for performing smoothing of pixels located at the boundaries of adjoining partial image regions on the basis of predetermined evaluation criteria, with respect to the prediction image obtained by the motion compensation predicting means; and

prediction residual coding means for coding the prediction residual signal obtained from the difference between the input image and the smoothed prediction image; wherein the smoothing means comprise

activity level setting means for determining the activity level of one or more of the adjoining partial image regions depending on whether respective coding modes for the adjoining partial images indicate inter coding or intra coding;

adaptive smoothing means for changing a number of pixels subjected to smoothing across a boundary between the partial image regions in accordance with the activity levels obtained by the activity level setting means and for then performing smoothing of said number of pixels; and

post-processing means for performing suppression processing of the results of smoothing by means of a threshold value established in accordance with the activity levels, wherein the differential values between the pixel values before the smoothing and the pixel values after the smoothing are controlled by means of the threshold value, wherein the threshold is the maximum permissible differential value, and in cases where the differential value between a pixel value before the smoothing and a pixel value after the smoothing is equal to or more than the threshold value, the pixel value after the smoothing is corrected to diminish the differential value in accordance with the size of the threshold value."

XIV. Claim 1 of the amended sixth auxiliary request reads as follows (additions to the sixth auxiliary request indicated by underlining, deletions by strike-through):

"An image coding apparatus, characterized by comprising:

motion compensation predicting means for generating a motion-compensated prediction image by detecting movement amounts in predetermined partial image region units of an input image;

smoothing means for performing smoothing of pixels located at the boundaries of adjoining partial image regions on the basis of predetermined evaluation criteria, with respect to the prediction image obtained by the motion compensation predicting means; and

prediction residual coding means for coding the prediction residual signal obtained from the difference between the input image and the smoothed prediction image; wherein the smoothing means perform the smoothing in accordance with a difference value for the movement amount between the adjoining partial image regions; and

wherein the smoothing means comprise

activity level setting means for determining the activity level of [deleted: one or more of] the adjoining partial image regions depending on whether respective coding modes for the adjoining partial image[deleted: s] regions indicate inter coding or intra coding;

adaptive smoothing means for changing a number of pixels subjected to smoothing across a boundary between the partial image regions in accordance with the activity levels obtained by the activity level setting means and for then performing smoothing of said number of pixels; and

post-processing means for performing suppression processing of the results of smoothing by means of a threshold value established in accordance with the activity levels, wherein the differential values between the pixel values before the smoothing and the pixel values after the smoothing are controlled by means of the threshold value, wherein the threshold is the maximum permissible differential value, and in cases where the differential value between a pixel value before the smoothing and a pixel value after the smoothing is equal to or more than the threshold value, the pixel value after the smoothing is corrected to diminish the differential value in accordance with the size of the threshold value."

XV. The independent claims of the seventh auxiliary request read as follows (amendments to claim 1 of the amended sixth auxiliary request are underlined):

"1. An image coding apparatus, characterized by comprising:

motion compensation predicting means for generating a motion-compensated prediction image by detecting movement amounts in predetermined partial image region units of an input image;

smoothing means for performing smoothing of pixels located at the boundaries of adjoining partial image regions on the basis of predetermined evaluation criteria, with respect to the prediction image obtained by the motion compensation predicting means, wherein the adjoining partial image regions comprise a central image region and neighboring image regions each sharing one boundary with the central image region; and

prediction residual coding means for coding the prediction residual signal obtained from the difference between the input image and the smoothed prediction image;

wherein the smoothing means perform the smoothing in accordance with a difference value for the movement amount between the adjoining partial image regions; and wherein the smoothing means comprise

activity level setting means for determining the activity level of the adjoining partial image regions depending on whether respective coding modes for the adjoining partial image regions indicate inter coding or intra coding, wherein the activity level setting means adjusts the activity level according to a first rule, if the central image region is intra coded, and a second, different rule, if the central image region is inter coded;

adaptive smoothing means for changing a number of pixels subjected to smoothing across a boundary between the partial image regions in accordance with the activity levels obtained by the activity level setting means and for then performing smoothing of said number of pixels; and

post-processing means for performing suppression processing of the results of smoothing by means of a threshold value established in accordance with the activity levels, wherein the differential values between the pixel values before the smoothing and the pixel values after the smoothing are controlled by means of the threshold value, wherein the threshold is the maximum permissible differential value, and in cases where the differential value between a pixel value before the smoothing and a pixel value after the smoothing is equal to or more than the threshold value, the pixel value after the smoothing is corrected to diminish the differential value in accordance with the size of the threshold value."

"5. An image decoding apparatus, characterized by comprising:

motion compensation predicting means for generating a motion-compensated prediction image by detecting movement amounts in predetermined partial image region units;

smoothing means for performing smoothing of pixels located at the boundaries of adjoining partial image regions on the basis of predetermined evaluation criteria, with respect to the prediction image obtained by the motion compensation predicting means; wherein the adjoining partial image regions comprise a central image region and neighboring image regions each sharing one boundary with the central image region;

prediction residual decoding means for decoding a prediction residual signal from the encoding side; and adding means for obtaining a decoded image by adding together a decoded prediction residual signal obtained by the prediction residual decoding means, and the smoothed prediction image;

wherein the smoothing means perform the smoothing in accordance with a difference value for the movement amount between the adjoining partial image regions; and wherein the smoothing means comprise

activity level setting means for determining the activity level of the adjoining partial image regions depending on whether respective coding modes for the adjoining partial image regions indicate inter coding or intra coding, wherein the activity level setting means adjusts the activity level according to a first rule, if the central image region is intra coded, and a second, different rule, if the central image region is inter coded;

adaptive smoothing means for changing a number of pixels subjected to smoothing across a boundary between the partial image regions in accordance with the activity levels obtained by the activity level setting means and for then performing smoothing of said number of pixels; and

post-processing means for performing suppression processing of the results of smoothing by means of a threshold value established in accordance with the activity levels, wherein the differential values between the pixel values before the smoothing and the pixel values after the smoothing are controlled by means of the threshold value, wherein the threshold is the maximum, permissible differential value, and in cases where the differential value between a pixel value before the smoothing and a pixel value after the smoothing is equal to or more than the threshold value, the pixel value after the smoothing is corrected to diminish the differential value in accordance with the size of the threshold value."

"9. An image coding method, comprising: a motion compensation predicting step of generating a motion-compensated prediction image by detecting movement amounts in predetermined partial image region units of an input image;

a smoothing step of performing smoothing of pixels located at the boundaries of adjoining partial image regions on the basis of predetermined evaluation criteria, with respect to the prediction image obtained by the motion compensation prediction step; wherein the adjoining partial image regions comprise a central image region and neighboring image regions each sharing one boundary with the central image region; and

a prediction residual coding step of coding the prediction residual signal obtained from the difference between the input image and the smoothed prediction image;

wherein the smoothing means perform the smoothing in accordance with a difference value for the movement amount between the adjoining partial image regions and

wherein the smoothing step

determines the activity level of the adjoining partial image regions depending on whether respective coding modes for the adjoining partial image regions indicate inter coding or intra coding, wherein the activity level is adjusted according to a first rule, if the central image region is intra coded, and a second, different rule, if the central image region is inter coded;

changing a number of pixels subjected to smoothing across a boundary between the partial image regions in accordance with the activity levels obtained and then performing smoothing of said number of pixels; and

performing suppression processing of the results of smoothing by means of a threshold value established in accordance with the activity levels, wherein the differential values between the pixel values before the smoothing and the pixel values after the smoothing are controlled by means of the threshold value, wherein the threshold is the maximum permissible differential value, and in cases where the differential value between a pixel value before the smoothing and a pixel value after the smoothing is equal to or more than the threshold value, the pixel value after the smoothing is corrected to diminish the differential value in accordance with the size of the threshold value."

"14. An image decoding method, characterized by comprising:

a motion compensation predicting step of generating a motion-compensated prediction image by detecting movement amounts in predetermined partial image region units;

a smoothing step of performing smoothing of pixels located at the boundaries of adjoining partial image regions on the basis of predetermined evaluation criteria, with respect to the prediction image obtained by the motion compensation predicting step; wherein the adjoining partial image regions comprise a central image region and neighboring image regions each sharing one boundary with the central image region; and

a prediction residual decoding step of decoding a prediction residual signal from the encoding side; and

an adding step of obtaining a decoded image by adding together a decoded prediction residual signal obtained by the prediction residual decoding step, and the smoothed prediction image;

wherein the smoothing means perform the smoothing in accordance with a difference value for the movement amount between the adjoining partial image regions; and

wherein the smoothing step

determines the activity level of the adjoining partial image regions depending on whether respective coding modes for the adjoining partial image regions indicate inter coding or intra coding, wherein the activity level is adjusted according to a first rule, if the central image region is intra coded, and a second, different rule, if the central image region is inter coded;

changing a number of pixels subjected to smoothing across a boundary between the partial image regions in accordance with the activity levels obtained and then performing smoothing of said number of pixels; and

performing suppression processing of the results of smoothing by means of a threshold value established in accordance with the activity levels,

wherein the differential values between the pixel values before the smoothing and the pixel values after the smoothing are controlled by means of the threshold value, wherein the threshold is the maximum permissible differential value, and in cases where the differential value between a pixel value before the smoothing and a pixel value after the smoothing is equal to or more than the threshold value, the pixel value after the smoothing is corrected to diminish the differential value in accordance with the size of the threshold value."

XVI. In the decision under appeal the examining division gave reasons why the independent claims of the fourth auxiliary request lacked clarity and inventive step. It argued that these reasons applied equally to the remaining requests (main request, first to third and fifth auxiliary requests), since the claims of these requests were less limited than those of the fourth auxiliary request (see decision under appeal, Reasons, section 1).

The examining division found that the wording "activity level between adjoining partial image regions" in claim 1 was unclear. In image coding, activity levels were normally defined within an image region, e.g. a macroblock, and not between blocks. It also argued that the wording "establishing the intensity of the smoothing on the basis of the activity level" did not specify how the activity level determined the intensity of the smoothing, and "performing smoothing of the pixels located at the boundaries" was unclear because it did not define sufficiently which pixels were subjected to smoothing (see decision under appeal, Reasons, points 3.1 to 3.3).

D1 was considered the closest prior art, which did not disclose the activity level setting means, the adaptive smoothing means and the post-processing means of claim 1 of the fourth auxiliary request. The technical problem solved by these distinguishing features was to prevent artificial boundaries of blocks from standing out among macroblocks of intra mode. The distinguishing features were disclosed in D4 as providing the same advantages as in the present application. D4 also referred to the same type of encoder as the one disclosed in D1. The skilled person would regard it as a normal design option to include the features of D4 in the apparatus described in D1 in order to solve the technical problem (see decision under appeal, Reasons, section 4).

XVII. The appellant's arguments, as far as they are relevant for the present decision, may be summarised as follows:

The wording "activity level between adjoining partial image regions" in claim 1 of the fourth auxiliary request was clear, because the activity level concerned the relationship between two blocks or "with the neighboring blocks" (see last paragraph of page 13 and page 16, third full paragraph). "Between adjoining blocks" was to be understood in the sense of "amongst adjoining blocks".

The specification of an "activity level of one or more of the adjoining partial image regions" in claim 1 of the sixth auxiliary request was not to be understood to mean that only one region was considered. Instead, the reference to adjoining regions implied that several regions were involved. The feature was disclosed on page 16, third full paragraph; page 10, the paragraph starting with "The MC procedure ..."; page 13, last paragraph and in Rule 2(1) to (5) (see pages 14 and 15) which provided a basis for the interpretation that the activity level(s) of other partial image region(s) could be determined differently than by evaluation of the coding mode. Furthermore, in an extreme case the image could be divided into two blocks with one common boundary. In that case the activity level could be determined for only one block, which necessarily took into account the second block.

D1 was the closest prior art with respect to the subject-matter of claim 1 of the amended sixth auxiliary request. The skilled person would not have combined D1 and D4, because these documents related to incompatible concealment algorithms. D4 disclosed smoothing of a decoded image, not of a predicted image as in D1. In addition, D1 related to a different problem resulting from the panning of a video camera (see column 4, line 63 to column 5, line 6).

In the oral proceedings the appellant agreed that the technical problem had been correctly formulated as how to prevent artificial boundaries of blocks from standing out among macroblocks of intra mode. The main teaching of D1 was that intra-loop filtering could be selectively turned off, based on considerations unrelated to those in D4 which was based on the distinction between intra or inter coding modes. Hence, the skilled person would not have combined D1 and D4. Even if such a combination had been considered, (s)he would not have arrived at the claimed solution.

Reasons for the Decision

1. The appeal is admissible.

The present application

2. The present invention relates to an image coding apparatus, image decoding apparatus and corresponding methods of encoding and decoding that alleviate artefacts resulting from block-based motion compensation. Motion compensation (MC) is effected based on blocks predicted from a previous image, for example. However, due to the fact that optimal prediction does not result in the same MC vector for all blocks, discontinuities are perceived at the boundaries of MC blocks. Discontinuities are particularly visible under circumstances of disparate motion vectors and for intra-coded blocks which are in general quantised at a low resolution (see pages 1 to 4, page 18, first paragraph and figures 21 and 22 of the application).

The present invention proposes using a smoothing filter on the motion-compensated prediction image. This filter selectively filters pixels located at the boundaries of the MC blocks after determining an activity level that depends on whether inter- or intra-coding mode has been used for the MC blocks and on the difference in motion vectors. The results of the smoothing undergo a post-processing operation such that effects of the smoothing are limited to a maximum permissible value (see pages 10 to 18 and figure 1).

Main request and first to fifth auxiliary requests

Clarity (Article 84 EPC 1973)

3. According to Article 84 EPC 1973, the claims shall define the matter for which protection is sought. They shall be clear and concise and be supported by the description.

3.1 Claim 1 of each of the main request and the first to fifth auxiliary requests refers to an "activity level between adjoining partial image regions".

3.2 The board agrees with the decision under appeal, which held that in image coding, activity levels are normally defined for blocks and not between blocks (see point XVI above), with the term "between blocks" being understood as "in the space that separates the blocks". This understanding of the term is also consistent with the description of the present application, which refers to "a block activity level S(X) in units of fixed blocks X" (see page 13, last paragraph). This definition implies that the activity level is associated with a block X and not with a relationship between two blocks X and Y. This interpretation is also supported by figure 6, see steps ST1 and ST6 and the paragraph bridging pages 15 and 16.

3.3 Also, the third full paragraph on page 16, which was relied on by the appellant to support its interpretation, is unambiguous in this respect. It specifies "In the filter processing below, processing is performed for two block boundaries and therefore, the larger value of the values of S(X) for the two blocks processed is used as the block activity level S." The passage clearly refers to "values of S(X) for the two blocks" (underlining added by the board), which implies that there are (at least) two values for two blocks and not a single value specifying a relationship between the blocks. The fact that the larger of the activity values is selected to control the filter processing is a different issue which relates to the subsequent use of the determined activity values.

The appellant argued in addition that "between adjoining blocks" was to be understood in the sense of "amongst adjoining blocks". According to this interpretation the activity level is determined in some unspecified way for a group of blocks, which is not consistent with the description of the present application (see the above cited passages).

4. It follows from the above that claim 1 of each of the main and first to fifth auxiliary requests does not meet the requirements of Article 84 EPC 1973.

Sixth auxiliary request

Added subject-matter, Article 123(2) EPC

5. Claim 1 of the sixth auxiliary request specifies "activity level setting means for determining the activity level of one or more of the adjoining partial image regions".

5.1 It was undisputed that the option of activity level setting means determining the activity level of one of the adjoining partial image regions was not explicitly disclosed in the application as filed. However, according to the appellant the application as filed implicitly provided a basis for this feature on page 10, the paragraph starting with "The MC procedure ..."; page 13, last paragraph; page 16, third full paragraph and pages 14 and 15, Rule 2(1) and (5).

5.2 The passage on page 10 only describes the division of macroblocks into rectangular image regions. The last paragraph on page 13 specifies that the "block activity level calculation section 125 determines a block activity level S(X) in units of fixed blocks X". S(X) is initialised to a value S0 over the whole area of the frame. The passage also refers to the flow of processing in figure 6, which shows a loop being carried out until "finished for all fixed blocks". Hence, these passages do not disclose the determination of an activity level for one block.

The third paragraph on page 16 concerns the influence of the activity level on the filtering of block boundaries. The strength of the filtering depends on the larger of the activity levels of the two blocks sharing the boundary. Hence, according to this passage two activity levels are determined for two neighbouring blocks.

The appellant argued that it followed from the latter passage that in an extreme case the image could be divided into two blocks with one common boundary. In that case the activity level could be determined for only one block, which necessarily took the other block into account. The board cannot agree. Even if the image were only composed of two MC blocks, it follows from the third paragraph on page 16 that the activity level would be determined for both blocks, with the larger of the two levels being chosen to determine the filtering strength.

Finally, both Rules 1 and 2 (see application, pages 13 to 15) are dependent on the coding mode, so that the coding mode is evaluated for all blocks.

5.3 Hence, claim 1 of the sixth auxiliary request contains subject-matter extending beyond the content of the application as filed (Article 123(2) EPC).

Amended sixth auxiliary request

Inventive step, Article 56 EPC 1973

6. It is common ground that D1 may be considered to be the closest prior art for the subject-matter of claim 1.

6.1 It is also undisputed that D1 (see figure 1 and column 7, line 40 to column 8, line 11) discloses the motion compensation predicting means, smoothing means and prediction residual coding means of claim 1 of the sixth auxiliary request. These means correspond to the general structure of the image coding apparatus of claim 1. D1 starts from the realisation that intra-loop filtering (i.e. smoothing) is ineffective if the motion vectors of adjacent blocks are substantially identical, which is the case if a picture is taken by panning a television camera. Hence, according to D1 the smoothing means perform the smoothing in accordance with a difference value for the movement amount between adjoining partial image regions, such that intra-loop filtering is switched off when adjacent blocks have substantially identical motion vectors and switched on if the motion vectors differ (see figure 1; column 4, line 63 to column 5, line 6; column 8, line 45 to column 9, line 7 and column 10, lines 32 to 45).

6.2 The subject-matter of claim 1 differs from D1 in the implementation of the smoothing means. In particular, the smoothing means of claim 1 contain

(a) an activity level setting means for determining the activity level of the adjoining partial image regions depending on whether respective coding modes for the adjoining partial image regions indicate inter coding or intra coding;

(b) adaptive smoothing means for changing a number of pixels subjected to smoothing across a boundary between the partial image regions in accordance with the activity levels obtained by the activity level setting means, and for then performing smoothing of said number of pixels; and

(c) post-processing means for performing suppression processing of the results of smoothing by means of a threshold value established in accordance with the activity levels, wherein the differential values between the pixel values before the smoothing and the pixel values after the smoothing are controlled by means of the threshold value, wherein the threshold is the maximum permissible differential value, and in cases where the differential value between a pixel value before the smoothing and a pixel value after the smoothing is equal to or more than the threshold value, the pixel value after the smoothing is corrected to diminish the differential value in accordance with the size of the threshold value.

6.3 The technical effect of these features is that suppression of artefacts in intra mode can be improved. The corresponding technical problem can be formulated as how to prevent artificial boundaries of blocks from standing out among macroblocks of intra mode.

6.4 D4 is a document drawn up by a standardisation committee concerned with video coding. It discloses a deblocking filter (i.e. a smoothing means) designed to solve the above technical problem. It discloses that the smoothing means have activity level setting means for determining the activity level in dependence on the coding mode (see chapter 4.5.3, the condition that "One or both of the adjacent macroblocks are intra coded"). It also discloses that the smoothing means are adaptively changing a number of pixels subjected to smoothing (according to chapter 4.5.3 three pixels to the left and to the right of a macroblock boundary in a luminance frame are filtered, "Otherwise filtering is done as described in the section above", i.e. zero to two pixels are filtered, see chapter 4.5.2). D4 also discloses the post-processing means of claim 1, wherein the maximum permissible differential value is controlled by means of the threshold value C (see bottom of chapter 4.5.2).

6.5 The board holds that the skilled person would have realised that the smoothing according to D1 could be further improved by incorporating the features of the deblocking filter of D4.

6.6 The appellant argued that the skilled person would not have combined D1 and D4, because these documents related to incompatible concealment algorithms. D4 disclosed smoothing of a reconstructed/decoded image and not of a predicted image as in D1 (see its chapter 4.5.1, first sentence).

D4 does not explicitly specify the location of its deblocking filter, apart from a reference to the reconstructed picture which is used to assign a strength value to each block. However, the board sees no obstacle preventing the skilled person from implementing the smoothing algorithm of D4 in the smoothing filter of D1. On the contrary, starting from D1 as the closest prior art the most straightforward solution for implementing the smoothing algorithm of D4 in D1 would have been to place its additional filtering at the same place as in D1, i.e. in the smoothing filter.

The appellant argued correctly that D1 related to an improvement which was effective for video sequences resulting from the panning of a video camera. In these situations the modified algorithm of D1 provides better image quality, because the smoothing filter is selectively switched off. However, the board regards the fact that the smoothing filter of D1 is adapted to different input sequences as showing that the skilled person was aware of shortcomings in the smoothing filter for certain input sequences and that these shortcomings could be alleviated by an adaptive filter. Filtering based on intra or inter coding mode is just a further adaptation which is effective under specific conditions and can be added to the existing filter. At least in circumstances in which this further adaptation is effective, a person skilled in the art would have added it in order to alleviate the shortcomings and so would have arrived at the subject-matter of claim 1.

6.7 As a result, the subject-matter of claim 1 would have been obvious to a skilled person in view of D1 and thus lacks inventive step (Article 56 EPC 1973).

Seventh auxiliary request

Added subject-matter, Article 123(2) EPC

7. Claim 1 of the seventh auxiliary request essentially differs from claim 1 of the fourth auxiliary request (whose compliance with Article 123(2) EPC was undisputed) in the following features:

(a) the adjoining partial image regions comprise a central image region and neighbouring image regions each sharing one boundary with the central image region;

(b) the smoothing means perform the smoothing in accordance with a difference value for the movement amount between the adjoining partial image regions;

(c) activity level setting means for determining the activity level of the adjoining partial image regions depending on whether respective coding modes for the adjoining partial image regions indicate inter coding or intra coding;

(d) the activity level setting means adjust the activity level according to a first rule, if the central image region is intra coded, and a second, different rule, if the central image region is inter coded;

(e) adaptive smoothing means for changing a number of pixels subjected to smoothing across a boundary between the partial image regions in accordance with the activity levels obtained by the activity level setting means, and for then performing smoothing of said number of pixels.

7.1 Feature (a) is based on figure 7 and page 13, last paragraph of the application. Feature (b) corresponds to claim 3 as originally filed. Features (c) and (d) are disclosed as the first embodiment of the invention, see in particular the passage starting on page 13, last paragraph to page 15. Feature (e) is based on claim 10 as originally filed and on figures 5 and 8, together with page 16, last two paragraphs and the first two paragraphs on page 17.

7.2 Corresponding amendments have been made to the other independent claims and - where necessary - dependent claims have been deleted.

7.3 Hence, the board is satisfied that the requirements of Article 123(2) EPC are fulfilled.

Clarity, Article 84 EPC 1973

8. The objected-to wording "activity level between adjoining partial image regions" in claim 1 has been amended to read "activity level of the adjoining partial image regions". In addition, according to the decision under appeal, the wording "establishing the intensity of the smoothing on the basis of the activity level" and "performing smoothing of the pixels located at the boundaries" was unclear because it did not define sufficiently which pixels were subjected to smoothing and how the intensity of the smoothing was determined (see point XVI above). Both wordings have been amended to further specify the smoothing operation.

8.1 Hence, the clarity objections of the decision under appeal have been successfully addressed. There is no other aspect under which the board considers the present set of claims to lack clarity.

8.2 It follows from the above that the claims of the seventh auxiliary request meet the requirements of Article 84 EPC 1973.

Inventive step, Article 56 EPC 1973

9. Compared with claim 1 of the amended sixth auxiliary request, claim 1 of the seventh auxiliary request contains the additional features (a) and (d), see point 7 above.

9.1 In D4 these features are not disclosed in the same context as in claim 1. D4 discloses changing the threshold of the post-processing means if the central image region (D4: C) is intra coded (see D4, section 4.5.1 and last paragraph of section 4.5.2). The activity level in claim 1 is considered to correspond to the activity parameters "ar" or "al" according to section 4.5.2 or alternatively to the combination of the conditions mentioned in section 4.5.3, including "one or both of the adjacent macroblocks being intra coded". Neither the one nor the other activity level is changed in dependence on whether the central image region is intra coded. Instead, in D4 the smoothing according to section 4.5.3 is only applied if adjacent macroblocks are intra coded. Hence, according to D4 the coding mode of the central image region is not taken into account for determining the activity level.

9.2 Based on the documents on file the board sees no reason why the person skilled in the art, starting from document D1 and adapting its teaching in view of D4, would have modified the algorithm of D4 in such a way as to arrive at the claimed subject-matter.

9.3 Hence, the subject-matter of claim 1 is considered to involve an inventive step (Article 56 EPC 1973). The same applies to independent claims 5, 9 and 14 which specify an image decoding apparatus, an image encoding method and an image decoding method corresponding to the image encoder of claim 1. Further claims 2 to 4, 6 to 8, 10 to 13 and 15 to 18 are dependent claims and, hence, also allowable.

Conclusion

10. The board finds that, taking into consideration the amendments made by the appellant during the appeal proceedings, the claims of the seventh auxiliary request meet the requirements of the EPC.| |

Order

For these reasons it is decided that:

1. The decision under appeal is set aside.

2. The case is remitted to the department of first instance with the order to grant a patent with the following claims and a description to be adapted thereto:

Claims No. 1 to 18 according to the Seventh Auxiliary Request filed during the oral proceedings of 8 June 2017.

Quick Navigation