Quick coding method for HEVC (high efficiency video coding)

1. A HEVC fast encoding method, which is characterized by comprising the following steps:

(1) the current in the high definition video to be processed for the current frame the frame;

(2) the current frame to be coded at present the largest encoding unit is defined as the current maximum coding unit;

(3) of the current frame in the video according to the high definition of the position of the maximum, and the current code unit when the position of the current frame, determining the current maximum coding unit of the largest all predicted coding unit, the current will be the greatest of all the prediction coding unit for coding unit to form the largest is defined as the set of the current maximum coding unit prediction set, and denoted as Ω, wherein Ω Ω is contained in or vantages of the largest encoding unit is L, LT, T, COL1 and COL2 at least one of in, COL1 to that of the current frame with the reference frame before the current maximum coding unit is the same as the coordinate position of the largest encoding unit, COL2 said reference frame to the current frame with the current maximum coding unit is the same as the coordinate position of the largest encoding unit, the current L the left side of the largest encoding unit adjacent the largest encoding unit, the current LT the largest encoding unit of a left upper side adjacent the largest encoding unit, the current T the upper edge of the largest encoding unit adjacent the largest encoding unit;

(4) the DR_pred the current maximum coding unit that the depth of the type value of the traverse interval, and then establishing DR_pred and the current maximum coding unit of the traverse section of the depth of the one-to-one correspondence relationship, wherein DR_pred is an integer, and 0     DR_pred ≤ 5;

(5) the complete Ω U= is judged whether it is equal to {COL1, COL2, L, LT, T}, if it is, step (6) is performed, otherwise, directly determining DR_pred value, then execute step (7);

(6) in each of the Ω the largest encoding unit to perform analysis of the quadtree structure type, determine in Ω each of the depth of the largest encoding unit of the type value of the traverse interval, then according to each of the Ω the largest encoding unit depth of the type value of the traverse interval, obtaining DR_pred value, then executing step (7);

(7) according to DR_pred value, determine the current depth of the largest encoding unit traverse interval, then the largest encoding unit according to the current traversing the section to the depth of the current traversing the largest encoding unit to perform coding, the traversal in the encoding process, the current largest encoding unit encoding unit to perform early cutting, the specific process:

(7)-1, the encoding unit of the greatest depth of the current traversing the section to be processed in a coding unit of a coding unit is defined as the current, the current coding unit is defined as the Macroblock right interlaminar lamination layer;

(7)-2, according to the current encoding unit of the encoding unit in deep value and the current position in the current layer, under determined that the current encoding unit of the encoding unit whether or not a layer of all early cutting can be carried out, for the specific process:

If the current encoding unit of the depth value is 0, it is judged that the current encoding unit of all the lower layer encoding unit does not carry out early cutting, then execute step (7)-4;

If the depth value of the current coding unit 1 or 2, then when the previous coded unit for the current layer in paragraph 1 or paragraph 2 or section 3 when the coding unit, the determination of the current encoding unit a layer of all coding unit does not carry out early cutting, then execute step (7)-4; and when the coding unit in the current layer 4 when the coding unit, the depth of the encoding unit judge whether the current value is less than the current maximum coding unit traversing the maximum value of the depth of the section, if it is, it is judged that the current encoding unit of the encoding units can be a layer of all early cutting, then execute step (7)-3, otherwise, as not, then execute step (7)-5;

If the depth value of the current coding unit 3, as not, then execute step (7)-5;

(7)-3, comparing the present encoding unit of the encoding unit a layer on the rate-distortion cost, and the current layer of the four coding unit in the rate-distortion cost and after with the threshold value a value obtained by determining, if the former is smaller than the latter, it is determined that the current encoding unit of the encoding unit to perform the next layer early cutting, then execute step (7)-5, if the former is greater than or equal to the latter, it is determined that the current coding unit to the next layer encoding unit does not carry out early cutting, then execute step (7)-4;

(7)-4, the depth of the encoding unit judge whether the current value is less than the current maximum coding unit traversing the maximum value of the depth of the section, if it is, then the current coding unit is further divided into 4 the same coding unit of the next layer, then the next layer encoding unit in a coding unit of a current to be processed as the current coding unit, and the current coding unit located as a Macroblock right interlaminar lamination of the layers, and then returning to step (7)-2 to continue, otherwise, directly executing step (7)-5;

(7)-5, the end of the current encoding unit encoding process, then judge whether the current maximum coding unit the depth of the traversal section whether all of the encoding unit, if it is, the current maximum encoding unit encoding process is finished, then execute step (8), otherwise, in the current layer and then judging the four coding unit whether the are, if in the current layer of the four coding unit are processed, on the current layer a layer of a coding unit in the encoding unit to be processed as the current coding unit, and the current coding unit located as a Macroblock right interlaminar lamination of the layers, then returning to step (7)-2 to continue, if in the current layer of the four coding unit has not been processed, then the current layer to be processed next in the encoding unit a as the current coding unit, then returning to step (7)-2 to continue execution;

(8) the current frame to be encoded next one the largest encoding unit as the current maximum coding unit, then returning to step (3) to continue, until all of the current frame are the largest coding of the coding unit;

(9) the next in the high definition video frame as the current frame the frame treats processing, then returning to step (2) to continue, until all in a high-definition TV video frame processing is completed, thus finishes the rapid high definition video coding.

2. A method for fast coding HEVC according to Claim 1, characterized in that the step (3) the current maximum coding unit in the prediction set for the determining process Ω:

If the current frame in a high-definition TV video Image in the set of I frame section 1 frame, is the largest when the coding unit is the current frame section 1 coding unit is the largest, for a vacuum Ω; when the coding unit is the largest section of the current frame 1 in addition to in paragraph 1 outside the largest of the coding unit for coding unit is the largest, Ω = {T}; when the coding unit is the largest section of the current frame 1 row, in addition to section 1 outside the largest of the coding unit for coding unit is the largest, Ω = {L}; dangdang the coding unit is the current in the frame eliminates section 1 and section 1 of the remaining th row is the largest when the coding unit, Ω = {L, LT, T};

If the current frame in a high-definition TV video Image in the set of non-I frame section 1 frame or the final 1 frame, is the largest when the coding unit is the current frame section 1 coding unit is the largest, Ω = {COL1, COL2}; when the coding unit is the largest section of the current frame 1 in addition to in paragraph 1 outside the largest of the coding unit for coding unit is the largest, Ω = {COL1, COL2, T}; when the coding unit is the largest section of the current frame 1 row, in addition to section 1 outside the largest of the coding unit for coding unit is the largest, Ω = {COL1, COL2, L}; dangdang the coding unit is the current in the frame eliminates section 1 and section 1 of the remaining th row is the largest when the coding unit, Ω = {COL1, COL2, L, LT, T};

If the current frame in a high-definition TV video Image in the set of non-I frame section 2 frame to 7 any in the frame, is the largest when the coding unit is the current frame section 1 coding unit is the largest, Ω = {COL1, COL2}; when the coding unit is the largest section of the current frame 1 in addition to in paragraph 1 outside the largest of the coding unit for coding unit is the largest, Ω = {COL1, COL2, T}; when the coding unit is the largest section of the current frame 1 row, in addition to section 1 outside the largest of the coding unit for coding unit is the largest, Ω = {COL1, COL2, L}; dangdang the coding unit is the current in the frame eliminates section 1 and section 1 of the remaining th row is the largest when the coding unit, Ω = {COL1, COL2, L, LT, T}.

3. A as in Claim 1 or Claim 2 HEVC fast coding method, which is characterized in that in the step (4) DR_pred and the current maximum coding unit of the traverse section of the depth of the one-to-one correspondence relationship is: when the DR_pred = 0 time, the greatest depth of the coding unit to traverse the interval [0, 1]; when the DR_pred = 1 time, the greatest depth of the coding unit to traverse the interval [0, 2]; when the DR_pred = 2 time, the greatest depth of the coding unit to traverse the interval [1, 2]; when the DR_pred = 3 time, the greatest depth of the coding unit to traverse the interval [1, 3]; when the DR_pred = 4 time, the greatest depth of the coding unit to traverse interval [2, 3]; when the DR_pred = 5 time, the greatest depth of the coding unit to traverse the interval [0, 3].

4. A according to Claim 3 HEVC fast coding method, which is characterized in that the step of determining (5) directly in DR_pred the value of the 5 ; (6) the step of the specific process is as follows:

(6)-1, is calculated in each Ω the largest encoding unit in the given reference depth values are respectively 0, 1, 2, 3 the absolute depth error and when, in the j Ω the largest section of the coding unit in the given value when the reference depth d of the absolute depth error and denoted as wherein 1     j≤ 5, depth_i Ω j in that the largest section of the coding unit in a i size is 4 × 4 of the depth value of the basic storage unit, depth_i value in the range of [0, 3], the value of the d 0 or 1 or 2 or 3, the symbol "| |" for taking the absolute value symbol;

(6)-2, the current Ω article to be treated a coding unit j is defined as the largest the current unit;

(6)-3, according to the value of the order of from small to large, the current unit in the given reference depth values are respectively 0, 1, 2, 3 of the absolute depth error and sequencing, and then selects the minimum value of the two absolute depth error and as the current unit of two candidate, when the ordering the rear section 2 and section 3 absolute depth error and the value is at the same time, a d value is selected for the absolute depth error and as the current unit among them of a plurality of candidate;

(6)-4, if the current unit to respectively two candidate and And It is determined that the depth of the current unit of the type value of the traverse interval 0 ; if the current unit to respectively two candidate and And It is determined that the depth of the current unit of the traverse section of the type value of 2 ; if the current unit to respectively two candidate and And It is determined that the depth of the current unit of the traverse of the type value is 3 ; if the current unit to respectively two candidate and And It is determined that the depth of the current unit of the type value of the traverse section 4 ; wherein Ω corresponding to j in that the largest section of the coding unit in the given reference depth value is 0, 1, 2, 3 the absolute depth error and the;

(6)-5, the j=j + 1, will Ω the next to be processed in the largest encoding unit as the current unit, then returning to step (6)-3 to continue, until Ω the five largest in the coding unit is processed, each of the obtained in Ω the largest encoding unit of the type value of the traverse section of the depth, wherein j=j + 1 "=" in the assignment symbol;

(6)-6, a candidate list of a type value, and the candidate list of the type value initialized to [1, 5, 5, 5, 5, 5, 3];

(6)-7, of five Ω for the maximum encoding unit of the traverse section of the depth of the alternative type value in the list of candidate type value in the middle of the five value, the type of the new candidate list value, then according to the value of the order of from small to large value to the new type of candidate list of the seven type value in order, then the type of the sorted list of candidate value as the median value in DR_pred value.

5. A according to Claim 4 HEVC fast coding method, which is characterized in that the steps (7)-3 in the depth value of the previous coded unit when to 1 time, the judging threshold value of 65 ; when the depth value of the previous coded unit 2 at the time, the judgment threshold value is set to 8.