CN103702127B - Motion estimation search range Forecasting Methodology based on motion vector dependency and system - Google Patents

Abstract

The present invention provides a motion estimation search range prediction method and system based on motion vector correlation. The method includes steps: S1, presetting the maximum search range SR _max and the minimum search range SR _min ; S2, calculating the prediction unit PU in different modes S3, according to the mode of the current PU, obtain the corresponding fitting coefficient; S4, according to the mode of the current PU, obtain the corresponding motion vector predictor and available space and time domain related candidate motion vectors; S5, calculate The cumulative difference between the motion vector predictor and each available spatial and temporal correlation candidate motion vector; S6, according to the fitting coefficient in S2, the motion vector predictor in S4, and the cumulative difference in S5, calculate the basic search range; S7. Determine the final search range SR _k according to the basic search range, the preset maximum search range SR _max and the minimum search range SR _min . The invention can dynamically adjust the search range, so as to achieve the purpose of reducing the calculation amount and shortening the search time.

Description

Motion estimation search range prediction method and system based on motion vector correlation

technical field

The invention relates to the technical field of video processing, in particular to a motion estimation search range prediction method and system based on motion vector correlation.

Background technique

With the development of technology, high-definition, high-frame-rate video has gradually become a new demand of people. High Efficiency Video Coding (HEVC, High Efficiency Video Coding) is a video coding developed by the Moving Pictures Experts Group (MPEG, Moving Pictures Experts Group) and the International Telecommunications Union Telecommunication Standardization Organization (ITU-T, ITU-T for ITU Telecommunications Standardization Sector). A new-generation video compression standard jointly developed by the Video Coding Joint Group (JCT-VC, Joint Collaborative Team Video Coding) jointly formed by the Expert Group (VCEG, Video Coding Experts Group). HEVC aims to compress the bit rate to 50% of the previous generation video coding standard H.264/MPEG-4AVC without reducing the video quality. However, the cost of bit rate reduction is a sharp increase in encoding complexity. Although there are many methods that can be used to reduce the huge amount of computation, there is still much work to be done to further speed up the encoding. HEVC adopts a coding tree unit (CTU, Coding Tree Unit) structure. Each CTU can contain one or more coding units (CU, Coding Unit), and each CU corresponds to a prediction unit (PU, Prediction Unit) and a transformation unit (TU, Transform Unit). During motion estimation (ME, Motion Estimation), the current PU needs to search for the best matching part in the coded reference frame.

At present, the way of PU to optimize the search process is mainly to obtain a better position by searching within a fixed search range. There are problems with this method:

1. Without considering the motion of the current PU itself, the search range cannot be reasonably limited according to the motion of the current PU;

2. Without considering the consistency of motion between the current PU and the surrounding CUs, the search range cannot be adaptively adjusted;

Contents of the invention

Aiming at the deficiencies of the prior art, the present invention provides an intra-frame encoding optimization method and system based on mode preprocessing, so that the encoding speed can be stably accelerated during the HEVC encoding operation.

To achieve the above object, the present invention is achieved through the following technical solutions:

A motion estimation search range prediction method based on motion vector correlation, comprising the following steps:

S1. Presetting the maximum search range SR _max and the minimum search range SR _min ;

S2. Calculate the fitting coefficients ah _k , av _k and bh _k , bv _k of the prediction unit PU in different modes, where the total number of modes of the prediction unit PU is Q, 0≤k<Q;

S3. Obtain corresponding fitting coefficients ah _k , av _k and bh _k , bv _k according to the mode k of the current PU;

S4. According to the mode k of the current PU, obtain the corresponding motion vector predictor (MVPh _k , MVPv _k ) and available spatial and temporal related candidate motion vectors (MVCh _ki , MVCv _ki ), where iCount _k represents the available spatial domain The total number of candidate motion vectors related to the time domain, 0≤i<iCount _k ;

S5. Calculate the cumulative difference (ΔMVTh _k , ΔMVTv _k ) between the motion vector predictor (MVPh _k , MVPv _k ) and each available spatial and temporal domain related candidate motion vector (MVCh _ki , MVCv _ki );

S6. According to the fitting coefficients ah _k , av _k and bh _k , bv _k in S2, the motion vector predictor (MVPh _k , MVPv _k ) in S4, and the cumulative difference (ΔMVTh _k , ΔMVTv _k ) in S5, calculate basic search scope SRBasic _k ;

S7. Determine the final search range SR _k according to the basic search range SRBasic _k in S6 and the maximum search range SR _max and minimum search range SR _min in S1.

Further, the step S2 includes:

S21. Determine all modes of the prediction unit PU;

S22. Obtain the available spatial and temporal motion-related candidate motion vectors (MVCh _ijk , MVCv _ijk ) of PU _jk in the test sequence, where 0≤k<Q, j represents the jth PU in the kth mode, and Q is The total number of modes of the prediction unit;

S23. Obtain the motion vector predictor (MVPh _jk , MVPv _jk ) and the actual motion vector (MVh _jk , MVv _jk ) of the PU _jk ;

S24. Calculate the difference between the motion vector predictor (MVPh _jk , MVPv _jk ) and each available space-domain and time-domain related candidate motion vector (MVCh _ijk , MVCv _ijk ), and sum after taking the absolute value:

${\mathrm{<span\; class="notranslate">\; \&Delta;MVTh</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; =</span>\underset{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 0</span>}}{\overset{{\mathrm{<span\; class="notranslate">\; iCount</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{<span\; class="notranslate">\; \&Sigma;</span>}}<span\; class="notranslate">\; |</span>{\mathrm{<span\; class="notranslate">\; MVCh</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}\mathrm{<span\; class="notranslate">\; j</span>}\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; MVPh</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; |</span>$

${\mathrm{<span\; class="notranslate">\; \&Delta;MVTv</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; =</span>\underset{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 0</span>}}{\overset{{\mathrm{<span\; class="notranslate">\; iCount</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{<span\; class="notranslate">\; \&Sigma;</span>}}<span\; class="notranslate">\; |</span>{\mathrm{<span\; class="notranslate">\; MVC</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}\mathrm{<span\; class="notranslate">\; j</span>}\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; MVPv</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; |</span>$

Among them, iCount _jk represents the number of available spatial domain and temporal domain related candidate motion vectors;

S25. Calculate the difference between the actual motion vector (MVh _jk , MVv _jk ) of PU _jk and the motion vector predictor (MVPh _jk , MVPv _jk ), and take the absolute value:

ΔMVh _jk ＝|MVh _jk -MVPh _jk |

ΔMVv _jk ＝|MVv _jk -MVPv _jk |

S26, construct the following two equations:

ah _k ×|MVPh _jk |+bh _k ×ΔMVTh _jk ＝ΔMVh _jk

av _k ×|MVPv _jk |+bv _k ×ΔMVTv _jk ＝ΔMVv _jk

Ah _k , av _k and bh _k , bv _k are calculated by least squares fitting algorithm.

Further, the step S5 includes:

Calculate the difference between the motion vector predictor (MVPh _k , MVPv _k ) and each available spatially and temporally correlated candidate motion vector (MVCh _ki , MVCv _ki ), taking the absolute value and summing:

${\mathrm{<span\; class="notranslate">\; \&Delta;MVTh</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; =</span>\underset{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 0</span>}}{\overset{{\mathrm{<span\; class="notranslate">\; iCount</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{<span\; class="notranslate">\; \&Sigma;</span>}}<span\; class="notranslate">\; |</span>{\mathrm{<span\; class="notranslate">\; MVCh</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; MVPh</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; |</span>$

${\mathrm{<span\; class="notranslate">\; \&Delta;MVTv</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; =</span>\underset{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 0</span>}}{\overset{{\mathrm{<span\; class="notranslate">\; iCount</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{<span\; class="notranslate">\; \&Sigma;</span>}}<span\; class="notranslate">\; |</span>{\mathrm{<span\; class="notranslate">\; MVC</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; MVPv</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; |</span>$

Among them, iCount _k represents the number of available spatial and temporal correlation candidate motion vectors.

Further, the step S6 includes:

Based on the fitting coefficients ah _k , av _k and bh _k , bv _k in S2, the motion vector predictor (MVPh _k , MVPv _k ) in S4, and the cumulative difference (ΔMVTh _k , ΔMVTv _k ) in S5, calculate the basic search scope:

SRh _k-basic ＝|ah _k ×ΔMVTh _k +bh _k ×|MVPh _k ||

SRv _k-basic =|av _k ×ΔMVTv _k +bv _k ×|MVPv _k ||.

SRBasic _k = SRh _k-basic + SRv _k-basic

Further, the step S7 includes:

Determine the final search range SR _k according to the basic search range SRBasic _k and the preset maximum search range SR _max and minimum search range SR _min as follows:

${\mathrm{<span\; class="notranslate">\; SR</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; =</span>\left\{\begin{array}{cc}{\mathrm{<span\; class="notranslate">\; SR</span>}}_{\mathrm{<span\; class="notranslate">\; min</span>}}<span\; class="notranslate">\; ,</span>& {\mathrm{<span\; class="notranslate">\; SRBasic</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; <</span>{\mathrm{<span\; class="notranslate">\; SR</span>}}_{\mathrm{<span\; class="notranslate">\; min</span>}}\\ {\mathrm{<span\; class="notranslate">\; SRBasic</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; ,</span>& {\mathrm{<span\; class="notranslate">\; SR</span>}}_{\mathrm{<span\; class="notranslate">\; min</span>}}<span\; class="notranslate">\; \&le;</span>{\mathrm{<span\; class="notranslate">\; SRBasic</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; \&le;</span>{\mathrm{<span\; class="notranslate">\; SR</span>}}_{\mathrm{<span\; class="notranslate">\; max</span>}}\\ {\mathrm{<span\; class="notranslate">\; SR</span>}}_{\mathrm{<span\; class="notranslate">\; max</span>}}<span\; class="notranslate">\; ,</span>& {\mathrm{<span\; class="notranslate">\; SRBasic</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; ></span>{\mathrm{<span\; class="notranslate">\; SR</span>}}_{\mathrm{<span\; class="notranslate">\; max</span>}}\end{array}\right.<span\; class="notranslate">\; .</span>$

Motion estimation search range prediction system based on motion vector correlation, the system includes:

The pre-search range setting module is used to pre-set the maximum search range SR _max and the minimum search range SR _min ;

The fitting coefficient calculation module is used to calculate the fitting coefficients ah _k , av _k and bh _k , bv _k of the prediction unit PU in different modes, wherein the total number of modes of the prediction unit PU is Q, and 0≤k<Q;

The fitting coefficient acquisition module of the corresponding mode is used to obtain the corresponding fitting coefficients ah _k , av _k and bh _k , bv _k according to the mode k of the current PU;

The predictor and related motion vector acquisition module of the corresponding mode is used to acquire the corresponding motion vector predictor (MVPh _k , MVPv _k ) and available spatial and temporal related candidate motion vectors (MVCh _ki , MVCv _ki ), wherein, iCount _k represents the total number of available spatial domain and time domain related candidate motion vectors, 0≤i<iCount _k ;

a correlation calculation module for calculating the cumulative difference (ΔMVTh _k , ΔMVTv _k ) between the motion vector predictor (MVPh _k , MVPv _k ) and each available spatial and temporal correlation candidate motion vector (MVCh _ki , MVCv _ki );

The basic search range calculation module is used to obtain the fitting coefficient ah _k , av _k and bh _k , bv _k in the module according to the fitting coefficient of the corresponding mode, the predictor of the corresponding mode and the motion vector prediction in the related motion vector acquisition module Sub (MVPh _k , MVPv _k ), the cumulative difference (ΔMVTh _k , ΔMVTv _k ) in the correlation calculation module, calculates the basic search range SRBasic _k ;

The final search range determination module is used to determine the final search range SR _k according to the basic search range SRBasic _k in the basic search range calculation module and the maximum search range SR _max and the minimum search range SR _min in the pre-search range setting module.

Further, the fitting coefficient calculation module includes:

A PU mode determination unit, configured to determine all modes of the prediction unit PU;

The related motion vector acquisition unit is used to acquire the available space domain and time domain motion related candidate motion vectors (MVCh _ijk , MVCv _ijk ) of PU _jk in the test sequence, 0≤k<Q, and j represents the kth mode in the kth mode j PUs, k is the mode of the current prediction unit;

A predictor and actual motion vector acquisition unit, configured to acquire the motion vector predictor (MVPh _jk , MVPv _jk ) and actual motion vector (MVh _jk , MVv _jk ) of the PU _jk ;

The correlation calculation unit is used to calculate the difference between the motion vector predictor (MVPh _jk , MVPv _jk ) and each available spatial and temporal correlation candidate motion vector (MVCh _ijk , MVCv _ijk ), and sum after taking the absolute value:

${\mathrm{<span\; class="notranslate">\; \&Delta;MVTh</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; =</span>\underset{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 0</span>}}{\overset{{\mathrm{<span\; class="notranslate">\; iCount</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{<span\; class="notranslate">\; \&Sigma;</span>}}<span\; class="notranslate">\; |</span>{\mathrm{<span\; class="notranslate">\; MVCh</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}\mathrm{<span\; class="notranslate">\; j</span>}\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; MVPh</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; |</span>$

${\mathrm{<span\; class="notranslate">\; \&Delta;MVTv</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; =</span>\underset{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 0</span>}}{\overset{{\mathrm{<span\; class="notranslate">\; iCount</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{<span\; class="notranslate">\; \&Sigma;</span>}}<span\; class="notranslate">\; |</span>{\mathrm{<span\; class="notranslate">\; MVC</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}\mathrm{<span\; class="notranslate">\; j</span>}\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; MVPv</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; |</span>$

Among them, iCount _jk represents the number of available spatial domain and temporal domain related candidate motion vectors;

The difference calculation unit is used to calculate the difference between the actual motion vector (MVh _jk , MVv _jk ) and the motion vector predictor (MVPh _jk , MVPv _jk ) of the PU _jk , and take the absolute value:

$\begin{array}{c}{\mathrm{<span\; class="notranslate">\; \&Delta;MVh</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; =</span><span\; class="notranslate">\; |</span>{\mathrm{<span\; class="notranslate">\; MVh</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; MVPh</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; |</span>\\ {\mathrm{<span\; class="notranslate">\; \&Delta;MVv</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; =</span><span\; class="notranslate">\; |</span>{\mathrm{<span\; class="notranslate">\; MVv</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; MVPv</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; |</span>\end{array}<span\; class="notranslate">\; ;</span>$

The coefficient fitting unit is used to construct the following two equation systems:

ah _k ×|MVPh _jk |+bh _k ×ΔMVTh _jk ＝ΔMVh _jk

av _k ×|MVPv _jk |+bv _k ×ΔMVTv _jk ＝ΔMVv _jk

Ah _k , av _k and bh _k , bv _k are calculated by least squares fitting algorithm.

Wherein, the correlation calculation module is used to calculate the difference between the motion vector predictor (MVPh _k , MVPv _k ) and each available spatial and temporal correlation candidate motion vector (MVCh _ki , MVCv _ki ), and obtain the absolute value and:

${\mathrm{<span\; class="notranslate">\; \&Delta;MVTh</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; =</span>\underset{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 0</span>}}{\overset{{\mathrm{<span\; class="notranslate">\; iCount</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{<span\; class="notranslate">\; \&Sigma;</span>}}<span\; class="notranslate">\; |</span>{\mathrm{<span\; class="notranslate">\; MVCh</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; MVPh</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; |</span>$

${\mathrm{<span\; class="notranslate">\; \&Delta;MVTv</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; =</span>\underset{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 0</span>}}{\overset{{\mathrm{<span\; class="notranslate">\; iCount</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{<span\; class="notranslate">\; \&Sigma;</span>}}<span\; class="notranslate">\; |</span>{\mathrm{<span\; class="notranslate">\; MVC</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; MVPv</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; |</span>$

Among them, iCount _k represents the number of available spatial and temporal correlation candidate motion vectors.

Further, the basic search range calculation module is used to acquire the fitting coefficients ah _k , av _k and bh _k , bv _k in the fitting coefficient acquisition module of the corresponding mode, and in the predictor and related motion vector acquisition module of the corresponding mode The motion vector predictor (MVPh _k , MVPv _k ), the cumulative difference (ΔMVTh _k , ΔMVTv _k ) in the correlation calculation module, calculates the basic search range SRBasic _k :

SRh _k-basic ＝|ah _k ×ΔMVTh _k +bh _k ×|MVPh _k ||

SRv _k-basic =|av _k ×ΔMVTv _k +bv _k ×|MVPv _k ||.

SRBasic _k = SRh _k-basic + SRv _k-basic

Further, the final search range determination module is used to determine the final search range SR _k according to the basic search range SRBasic _k and the preset maximum search range SR _max and minimum search range SR _min :

${\mathrm{<span\; class="notranslate">\; SR</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; =</span>\left\{\begin{array}{cc}{\mathrm{<span\; class="notranslate">\; SR</span>}}_{\mathrm{<span\; class="notranslate">\; min</span>}}<span\; class="notranslate">\; ,</span>& {\mathrm{<span\; class="notranslate">\; SRBasic</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; <</span>{\mathrm{<span\; class="notranslate">\; SR</span>}}_{\mathrm{<span\; class="notranslate">\; min</span>}}\\ {\mathrm{<span\; class="notranslate">\; SRBasic</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; ,</span>& {\mathrm{<span\; class="notranslate">\; SR</span>}}_{\mathrm{<span\; class="notranslate">\; min</span>}}<span\; class="notranslate">\; \&le;</span>{\mathrm{<span\; class="notranslate">\; SRBasic</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; \&le;</span>{\mathrm{<span\; class="notranslate">\; SR</span>}}_{\mathrm{<span\; class="notranslate">\; max</span>}}\\ {\mathrm{<span\; class="notranslate">\; SR</span>}}_{\mathrm{<span\; class="notranslate">\; max</span>}}<span\; class="notranslate">\; ,</span>& {\mathrm{<span\; class="notranslate">\; SRBasic</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; ></span>{\mathrm{<span\; class="notranslate">\; SR</span>}}_{\mathrm{<span\; class="notranslate">\; max</span>}}\end{array}\right.<span\; class="notranslate">\; .</span>$

The present invention has following beneficial effect:

The present invention utilizes the motion vector predictor (MVPh _k , MVPv _k ) of the current PU, the available spatial and temporal correlation candidate motion vectors (MVCh _ki , MVCv _ki ), fitting coefficients ah _k , av _k and bh _k , bv _k dynamically adjusts the basic search range SRBasic _k , and finally the final search range SR _k is jointly determined by the preset maximum search range SR _max , the minimum search range SR _min and the dynamically adjusted basic search range SRBasic _k . Since the present invention utilizes the joint candidate motion vectors in space and time domains and the motion vector predictor of the current PU to predict the motion estimation search range of the current PU, it can dynamically adjust the search range in real time according to the actual situation, thereby reducing the amount of calculation. It also reduces search time.

Description of drawings

In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following will briefly introduce the drawings that need to be used in the description of the embodiments or the prior art. Obviously, the accompanying drawings in the following description are For some embodiments of the present invention, those skilled in the art can also obtain other drawings based on these drawings without creative work.

FIG. 1 is a flowchart of a motion estimation search range prediction method based on motion vector correlation in Embodiment 1 of the present invention;

2 is a flow chart of a motion estimation search range prediction method based on motion vector correlation in Embodiment 2 of the present invention;

FIG. 3 is a schematic diagram of relative positions of candidate motion vectors related to space domain described in Embodiment 2 of the present invention;

FIG. 4 is a schematic diagram of relative positions of candidate motion vectors related to time domain described in Embodiment 2 of the present invention;

Fig. 5 is a schematic structural diagram of a motion estimation search range prediction system based on motion vector correlation in Embodiment 3 of the present invention.

detailed description

In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments It is a part of embodiments of the present invention, but not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by persons of ordinary skill in the art without creative efforts fall within the protection scope of the present invention.

Example 1

Embodiment 1 of the present invention proposes a motion estimation search range prediction method based on motion vector correlation, as shown in FIG. 1 , including the following steps:

Step 101: Preset the maximum search range SR _max and the minimum search range SR _min .

Step 102: Calculate the fitting coefficients ah _k , av _k and bh _k , bv _k of the prediction unit PU in different modes, where the total number of modes of the prediction unit PU is Q, and 0≤k<Q.

Step 103: Obtain the corresponding fitting coefficients ah _k , av _k and bh _k , bv _k according to the mode k of the current PU.

Step 104: According to the mode k of the current PU, obtain the corresponding motion vector predictor (MVPh _k , MVPv _k ) and available spatial and temporal related candidate motion vectors (MVCh _ki , MVCv _ki ), where iCount _k represents the available The total number of candidate motion vectors related to space domain and time domain, 0≤i<iCount _k .

Step 105: Calculate the cumulative difference (ΔMVTh _k , ΔMVTv _k ) of the motion vector predictor (MVPh _k , MVPv _k ) and each available spatially and temporally correlated candidate motion vector (MVCh _ki , MVCv _ki ).

Step 106: According to the fitting coefficients ah _k , av _k and bh _k , bv _k in step 102, the motion vector predictor (MVPh _k , MVPv _k ) in step 104, the accumulated difference (ΔMVTh _k , ΔMVTv _k ), calculate the basic search range SRBasic _k .

Step 107: Determine the final search range SR _k according to the basic search range SRBasic _k and the preset maximum search range SR _max and minimum search range SR _min .

It can be seen that the embodiment of the present invention utilizes the motion vector predictor (MVPh _k , MVPv _k ) of the current PU, the available spatial and temporal correlation candidate motion vectors (MVCh _ki , MVCv _ki ), fitting coefficients ah _k , av _k and bh _k , bv _k dynamically adjust the basic search range SRBasic _k , and finally the final search range SR _k is jointly determined by the preset maximum search range SR _max , the minimum search range SR _min and the dynamically adjusted basic search range SRBasic _k . From the above description, it can be seen that since the embodiment of the present invention uses the relevant candidate motion vectors in the space domain and the time domain to predict the motion estimation search range of the current PU, the search range can be dynamically adjusted in real time according to the actual situation, thereby achieving the goal of reducing the amount of calculation. Purpose.

Example 2

The implementation process of a preferred embodiment of the present invention will be described in more detail through Embodiment 2 below. Referring to Figure 2, the process includes the following steps:

Step 201: Preset the maximum search range SR _max and the minimum search range SR _min .

In this step, the value of SR _max can refer to the fixed search range of the prior art. For example, if the fixed search range is set to 64 before using the present invention for optimization, then the SR _max here is also set to 64; because When making predictions, the search range that may be obtained is very small and close to zero. If the predicted search range is used directly, the search result may not be good. At this time, a minimum search range SR _min is used to limit the search range, which can be obtained through the limited SR _min Appropriately increase the search range to obtain a more stable result. Generally, taking 4 for SR _min is a more appropriate value.

Step 202: Determine all modes of the prediction unit PU.

In this step, determine the modes of all different prediction units (PU, Prediction Unit): refer to the HM software, for example, for a 64x64 CTU, the mode Q=20 of the PU is respectively:

4×8,4×16,4×8,8×8,8×16,

8×32,12×16,16×4,16×8,16×12,

16×16,16×32,24×32,32×8,32×16,

32×24, 32×32, 32×64, 64×32, 64×64.

Since the motion conditions of the areas included in the PUs of different modes are different, better results can be obtained by using different fitting coefficients for the PUs of different modes.

Step 203: Obtain the available spatial and temporal motion-related candidate motion vectors (MVCh _ijk , MVCv _ijk ) of PU _jk in the test sequence, where 0≤k<Q, j represents the jth PU in the kth mode, Q is the total number of modes of the prediction unit.

In this step, an appropriate test sequence is determined, and candidate motion vectors (MVCh _ijk , MVCv _ijk ) related to the spatial domain and time domain motion available to PU _jk in different modes in the test sequence are obtained, including the left side of the current PU _jk spatial domain (L), The motion vectors of the lower left (LD), upper (U), upper right (RU) and upper left (LU) CUs (as shown in Figure 3) and 1 of the lower right (RD) or center (C) in the time domain motion vectors of a CU (as shown in Figure 4), where the priority of the lower right in the time domain motion vector is higher than that of the center, and record the number iCount _jk of the candidate motion vectors related to the space domain and time domain motion, where j represents the first j PUs, 0≤k<Q means the PU of the kth mode. During the implementation process, due to the different positions of the PU, the number iCount _jk of the actually available candidate motion vectors is also different. After the available candidate motion vectors are obtained, they are stored in front of the storage location in sequence, and the number iCount _jk of the final available candidate motion vectors will be recorded at the same time; the unusable candidate motion vectors are set to (0,0), and stored in the storage location later. Since only the previous iCount _jk candidate motion vectors are used in the calculation, the candidate motion vectors stored in the back are not used, so no matter how many motion vectors are actually obtained, the method can be processed correctly.

Step 204: Obtain the motion vector predictor (MVPh _jk , MVPv _jk ) and the actual motion vector (MVh _jk , MVv _jk ) of the PU _jk .

In this step, the HEVC test model HM will center on the motion vector predictor and search for the best matching position within the range of SR _max . This best position is the actual motion vector. This step is actually the process of motion estimation: before motion estimation, only the motion vector predictor is known, the actual motion vector is obtained through the motion vector predictor, and these two sets of values are recorded.

Step 205: Calculate the accumulated difference between the motion vector predictor (MVPh _jk , MVPv _jk ) and each available spatial and temporal correlation candidate motion vector (MVCh _ijk , MVCv _ijk ).

In this step, solve as follows:

${\mathrm{<span\; class="notranslate">\; \&Delta;MVTh</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; =</span>\underset{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 0</span>}}{\overset{{\mathrm{<span\; class="notranslate">\; iCount</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{<span\; class="notranslate">\; \&Sigma;</span>}}<span\; class="notranslate">\; |</span>{\mathrm{<span\; class="notranslate">\; MVCh</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}\mathrm{<span\; class="notranslate">\; j</span>}\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; MVPh</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; |</span>$

${\mathrm{<span\; class="notranslate">\; \&Delta;MVTv</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; =</span>\underset{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 0</span>}}{\overset{{\mathrm{<span\; class="notranslate">\; iCount</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{<span\; class="notranslate">\; \&Sigma;</span>}}<span\; class="notranslate">\; |</span>{\mathrm{<span\; class="notranslate">\; MVC</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}\mathrm{<span\; class="notranslate">\; j</span>}\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; MVPv</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; |</span>$

Step 206: Calculate the difference between the actual motion vector (MVh _jk , MVv _jk ) of PU _jk and the motion vector predictor (MVPh _jk , MVPv _jk ), and take the absolute value.

In this step, solve as follows:

ΔMVh _jk ＝|MVh _jk -MVPh _jk |

ΔMVv _jk ＝|MVv _jk -MVPv _jk |

Step 207: Construct a system of equations, and use the least squares fitting algorithm to obtain fitting coefficients ah _k , av _k and bh _k , bv _k .

In this step, the following two systems of equations are constructed:

ah _k ×|MVPh _jk |+bh _k ×ΔMVTh _jk ＝ΔMVh _jk

av _k ×|MVPv _jk |+bv _k ×ΔMVTv _jk ＝ΔMVv _jk

Where k represents different PU modes, j represents all the values belonging to mode k, ah _k , av _k and bh _k , bv _k are calculated by using the least squares fitting algorithm.

Step 208: According to the mode k of the current PU, obtain the corresponding motion vector predictor (MVPh _k , MVPv _k ) and available spatial and temporal related candidate motion vectors (MVCh _ki , MVCv _ki ), where iCount _k represents the available The total number of candidate motion vectors related to space domain and time domain, 0≤i<iCount _k .

In this step, the motion vector predictor (MVPh _k , MVPv _k ) of the current PU and the available spatial and temporal related candidate motion vectors (MVCh _ki , MVCv _ki ) of the current PU are obtained, including the left (L), lower left ( LD), upper (U), upper right (RU), and upper left (LU) motion vectors of 5 CUs (as shown in Figure 3) and 1 CU in the time domain lower right (RD) or center (C) motion vector (as shown in Figure 4), where the priority of the lower right in the time domain motion vector is higher than that of the center, where i represents one of all available spatial and temporal domain related candidate motion vectors, and records the motion vector number iCount _k ;

Step 209: Calculate the cumulative difference (ΔMVTh _k , ΔMVTv _k ) of the motion vector predictor (MVPh _k , MVPv _k ) and each available spatially and temporally correlated candidate motion vector (MVCh _ki , MVCv _ki ).

In this step, the cumulative difference (ΔMVTh _k , ΔMVTv _k ) is solved as follows:

${\mathrm{<span\; class="notranslate">\; \&Delta;MVTh</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; =</span>\underset{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 0</span>}}{\overset{{\mathrm{<span\; class="notranslate">\; iCount</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{<span\; class="notranslate">\; \&Sigma;</span>}}<span\; class="notranslate">\; |</span>{\mathrm{<span\; class="notranslate">\; MVCh</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; MVPh</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; |</span>$

${\mathrm{<span\; class="notranslate">\; \&Delta;MVTv</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; =</span>\underset{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 0</span>}}{\overset{{\mathrm{<span\; class="notranslate">\; iCount</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{<span\; class="notranslate">\; \&Sigma;</span>}}<span\; class="notranslate">\; |</span>{\mathrm{<span\; class="notranslate">\; MVC</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; MVPv</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; |</span>$

Step 210: According to the fitting coefficients ah _k , av _k and bh _k , bv _k in step 207, the motion vector predictor (MVPh _k , MVPv _k ) in step 208, the cumulative difference (ΔMVTh _k , ΔMVTv _k ), calculate the basic search range SRBasic _k .

In this step, the basic search range SRBasic _k is solved as follows:

SRh _k-basic ＝|ah _k ×ΔMVTh _k +bh _k ×|MVPh _k ||

SRv _k-basic ＝|av _k ×ΔMVTv _k +bv _k ×|MVPv _k ||

SRBasic _k = SRh _k-basic + SRv _k-basic

(ΔMVTh _k , ΔMVTv _k ) represents the motion correlation between the current PU and the CU related to the space domain and the time domain: if both values are relatively small, it means that the motion of the current PU is consistent with that of the CU related to the space domain and the time domain. Search The range should be smaller; otherwise, it means that their motions are not consistent, and the search range should be larger. (MVPh _k , MVPv _k ) indicates the possible motion of the current PU: the two values are relatively small, indicating that the current PU itself is moving smoothly or there is no motion in the background, then the search range can be appropriately reduced; otherwise, the current PU The movement itself is more intense, so the search range should be larger.

Step 211: Determine the final search range SR _k according to the basic search range SRBasic _k and the preset maximum search range SR _max and minimum search range SR _min .

In this step, the final search range SR _k is solved in the following manner:

${\mathrm{<span\; class="notranslate">\; SR</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; =</span>\left\{\begin{array}{cc}{\mathrm{<span\; class="notranslate">\; SR</span>}}_{\mathrm{<span\; class="notranslate">\; min</span>}}<span\; class="notranslate">\; ,</span>& {\mathrm{<span\; class="notranslate">\; SRBasic</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; <</span>{\mathrm{<span\; class="notranslate">\; SR</span>}}_{\mathrm{<span\; class="notranslate">\; min</span>}}\\ {\mathrm{<span\; class="notranslate">\; SRBasic</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; ,</span>& {\mathrm{<span\; class="notranslate">\; SR</span>}}_{\mathrm{<span\; class="notranslate">\; min</span>}}<span\; class="notranslate">\; \&le;</span>{\mathrm{<span\; class="notranslate">\; SRBasic</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; \&le;</span>{\mathrm{<span\; class="notranslate">\; SR</span>}}_{\mathrm{<span\; class="notranslate">\; max</span>}}\\ {\mathrm{<span\; class="notranslate">\; SR</span>}}_{\mathrm{<span\; class="notranslate">\; max</span>}}<span\; class="notranslate">\; ,</span>& {\mathrm{<span\; class="notranslate">\; SRBasic</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; ></span>{\mathrm{<span\; class="notranslate">\; SR</span>}}_{\mathrm{<span\; class="notranslate">\; max</span>}}\end{array}\right.$

By adjusting the size of the SR in real time, the time for integer point search can be effectively reduced. Test results show that this method can reduce the search time of integer points by 20% to 55%, and at the same time ensure that the signal distortion rate is below 0.08dB.

In the embodiment of the present invention, by using the motion vector predictor (MVPh _k , MVPv _k ) of the current PU, the available spatial and temporal correlation candidate motion vectors (MVCh _ki , MVCv _ki ), fitting coefficients ah _k , av _k and bh _k ,bv _k dynamically adjusts the basic search range SRBasic _k , and finally the final search range SR _k is jointly determined by the preset maximum search range SR _max , the minimum search range SR _min and the dynamically adjusted basic search range SRBasic _k . From the above description, it can be seen that since the embodiment of the present invention uses the relevant candidate motion vectors in the space domain and the time domain to predict the motion estimation search range of the current PU, the search range can be dynamically adjusted in real time according to the actual situation, thereby reducing the amount of calculation and The purpose of shortening the search time.

Example 3

Embodiment 3 of the present invention also proposes a motion estimation search range prediction system based on motion vector correlation, see FIG. 5, the system includes:

Search range preset module 501, used to preset the maximum search range SR _max and the minimum search range SR _min ;

The fitting coefficient calculation module 502 is used to calculate the fitting coefficients ah _k , av _k and bh _k , bv _k of the prediction unit PU in different modes, wherein the total number of modes of the prediction unit PU is Q, 0≤k<Q;

The fitting coefficient acquisition module 503 of the corresponding mode is used to acquire the corresponding fitting coefficients ah _k , av _k and bh _k , bv _k according to the mode k of the current PU;

The predictor and related motion vector acquisition module 504 of the corresponding mode is used to acquire the corresponding motion vector predictor (MVPh _k , MVPv _k ) and available spatial and temporal related candidate motion vectors (MVCh _ki , MVCv _ki ), wherein, iCount _k represents the total number of available space and time domain related candidate motion vectors, 0≤i<iCount _k ;

The correlation calculation module 505 is used to calculate the cumulative difference (ΔMVTh _k , ΔMVTv _k ) between the motion vector predictor (MVPh _k , MVPv _k ) and each available spatial and temporal correlation candidate motion vector (MVCh _ki , MVCv _ki ) ;

The basic search range calculation module 506 is used to obtain the fitting coefficient ah _k , av _k and bh _k , bv _k in the module according to the fitting coefficient of the corresponding mode, and obtain the motion vector in the module according to the predictor and the related motion vector of the corresponding mode The predictor (MVPh _k , MVPv _k ), the cumulative difference (ΔMVTh _k , ΔMVTv _k ) in the correlation calculation module, calculates the basic search range SRBasic _k ;

The final search range determination module 507 is used to determine the final search range SR _k according to the basic search range SRBasic _k in the basic search range calculation module and the maximum search range SR _max and the minimum search range SR _min in the pre-search range setting module.

Wherein, the fitting coefficient calculation module 502 includes:

The PU mode determination unit 5020 is configured to determine all modes of the prediction unit PU:

In this unit, refer to the HM software, for example, for a 64x64 CTU, the PU mode Q=20, respectively:

4×8,4×16,4×8,8×8,8×16,

8×32,12×16,16×4,16×8,16×12,

16×16,16×32,24×32,32×8,32×16,

32×24, 32×32, 32×64, 64×32, 64×64.

Since the movement conditions of the areas included in the PUs of different modes are different, better results can be obtained by using different fitting coefficients for the PUs of different modes.

The related motion vector acquisition unit 5021 is used to acquire the available space domain and time domain motion related candidate motion vectors (MVCh _ijk , MVCv _ijk ) of PU _jk in the test sequence, where 0≤k<Q, and j represents the motion vector in the kth mode For the jth PU, Q is the total number of modes of the prediction unit;

In this unit, first determine the appropriate test sequence, and then obtain the candidate motion vectors (MVCh _ijk , MVCv _ijk ) related to the spatial and temporal motions available to PU _jk of different modes in the test sequence, including the left side of the current PU _jk spatial domain (L ), lower left (LD), upper (U), upper right (RU) and upper left (LU) 5 CU motion vectors (as shown in Figure 3) and time domain lower right (RD) or center (C) in both The motion vector of 1 CU (as shown in Figure 4), where the priority of the lower right in the time domain motion vector is higher than that of the center, and record the number of candidate motion vectors iCount _jk related to spatial and temporal motion, where j Indicates the jth PU, and 0≤k<Q indicates the kth mode of the PU. During the implementation process, due to the different positions of the PU, the number iCount _jk of the actually available candidate motion vectors is also different. After the available candidate motion vectors are obtained, they are stored in front of the storage location in sequence, and the number iCount _jk of the final available candidate motion vectors will be recorded at the same time; the unusable candidate motion vectors are set to (0,0), and stored in the storage location later. Since only the previous iCount _jk candidate motion vectors are used in the calculation, the candidate motion vectors stored in the back are not used, so no matter how many motion vectors are actually obtained, the method can be processed correctly.

A predictor and actual motion vector acquisition unit 5022, configured to acquire the motion vector predictor (MVPh _jk , MVPv _jk ) and actual motion vector (MVh _jk , MVv _jk ) of the PU _jk ;

The correlation calculation unit 5023 is used to calculate the difference between the motion vector predictor (MVPh _jk , MVPv _jk ) and each available space-domain and time-domain correlation candidate motion vector (MVCh _ijk , MVCv _ijk ), and sum after taking the absolute value:

${\mathrm{<span\; class="notranslate">\; \&Delta;MVTh</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; =</span>\underset{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 0</span>}}{\overset{{\mathrm{<span\; class="notranslate">\; iCount</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{<span\; class="notranslate">\; \&Sigma;</span>}}<span\; class="notranslate">\; |</span>{\mathrm{<span\; class="notranslate">\; MVCh</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}\mathrm{<span\; class="notranslate">\; j</span>}\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; MVPh</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; |</span>$

${\mathrm{<span\; class="notranslate">\; \&Delta;MVTv</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; =</span>\underset{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 0</span>}}{\overset{{\mathrm{<span\; class="notranslate">\; iCount</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{<span\; class="notranslate">\; \&Sigma;</span>}}<span\; class="notranslate">\; |</span>{\mathrm{<span\; class="notranslate">\; MVC</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}\mathrm{<span\; class="notranslate">\; j</span>}\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; MVPv</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; |</span>$

Among them, iCount _jk represents the number of available spatial domain and temporal domain related candidate motion vectors;

The difference calculation unit 5024 is used to calculate the difference between the actual motion vector (MVh _jk , MVv _jk ) of PU _jk and the motion vector predictor (MVPh _jk , MVPv _jk ), and take the absolute value:

$\begin{array}{c}{\mathrm{<span\; class="notranslate">\; \&Delta;MVh</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; =</span><span\; class="notranslate">\; |</span>{\mathrm{<span\; class="notranslate">\; MVh</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; MVPh</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; |</span>\\ {\mathrm{<span\; class="notranslate">\; \&Delta;MVv</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; =</span><span\; class="notranslate">\; |</span>{\mathrm{<span\; class="notranslate">\; MVv</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; MVPv</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; |</span>\end{array}<span\; class="notranslate">\; ;</span>$

The coefficient fitting unit 5025 is used to construct the following two equations:

ah _k ×|MVPh _jk |+bh _k ×ΔMVTh _jk ＝ΔMVh _jk

av _k ×|MVPv _jk |+bv _k ×ΔMVTv _jk ＝ΔMVv _jk

Ah _k , av _k and bh _k , bv _k are calculated by least squares fitting algorithm.

Wherein, the correlation calculation module 505 is used to calculate the difference between the motion vector predictor (MVPh _k , MVPv _k ) and each available spatial and temporal correlation candidate motion vector (MVCh _ki , MVCv _ki ), after taking the absolute value Summing:

${\mathrm{<span\; class="notranslate">\; \&Delta;MVTh</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; =</span>\underset{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 0</span>}}{\overset{{\mathrm{<span\; class="notranslate">\; iCount</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{<span\; class="notranslate">\; \&Sigma;</span>}}<span\; class="notranslate">\; |</span>{\mathrm{<span\; class="notranslate">\; MVCh</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; MVPh</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; |</span>$

${\mathrm{<span\; class="notranslate">\; \&Delta;MVTv</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; =</span>\underset{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 0</span>}}{\overset{{\mathrm{<span\; class="notranslate">\; iCount</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{<span\; class="notranslate">\; \&Sigma;</span>}}<span\; class="notranslate">\; |</span>{\mathrm{<span\; class="notranslate">\; MVC</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; MVPv</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; |</span>$

Among them, iCount _k represents the number of available spatial and temporal correlation candidate motion vectors.

Wherein, the basic search range calculation module 506 is used in the predictor and related motion vector acquisition module of the corresponding mode according to the fitting coefficient ah _k , av _k and bh _k , bv _k in the fitting coefficient acquisition module of the corresponding mode The motion vector predictor (MVPh _k , MVPv _k ), the cumulative difference (ΔMVTh _k , ΔMVTv _k ) in the correlation calculation module, calculates the basic search range SRBasic _k :

SRh _k-basic ＝|ah _k ×ΔMVTh _k +bh _k ×|MVPh _k ||

SRv _k-basic =|av _k ×ΔMVTv _k +bv _k ×|MVPv _k ||.

SRBasic _k = SRh _k-basic + SRv _k-basic

Wherein, the final search range determination module 507 is used to determine the final search range SR _k according to the basic search range SRBasic _k and the preset maximum search range SR _max and minimum search range SR _min :

${\mathrm{<span\; class="notranslate">\; SR</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; =</span>\left\{\begin{array}{cc}{\mathrm{<span\; class="notranslate">\; SR</span>}}_{\mathrm{<span\; class="notranslate">\; min</span>}}<span\; class="notranslate">\; ,</span>& {\mathrm{<span\; class="notranslate">\; SRBasic</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; <</span>{\mathrm{<span\; class="notranslate">\; SR</span>}}_{\mathrm{<span\; class="notranslate">\; min</span>}}\\ {\mathrm{<span\; class="notranslate">\; SRBasic</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; ,</span>& {\mathrm{<span\; class="notranslate">\; SR</span>}}_{\mathrm{<span\; class="notranslate">\; min</span>}}<span\; class="notranslate">\; \&le;</span>{\mathrm{<span\; class="notranslate">\; SRBasic</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; \&le;</span>{\mathrm{<span\; class="notranslate">\; SR</span>}}_{\mathrm{<span\; class="notranslate">\; max</span>}}\\ {\mathrm{<span\; class="notranslate">\; SR</span>}}_{\mathrm{<span\; class="notranslate">\; max</span>}}<span\; class="notranslate">\; ,</span>& {\mathrm{<span\; class="notranslate">\; SRBasic</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; ></span>{\mathrm{<span\; class="notranslate">\; SR</span>}}_{\mathrm{<span\; class="notranslate">\; max</span>}}\end{array}\right.<span\; class="notranslate">\; .</span>$

The system described in the embodiment of the present invention uses the motion vector predictor (MVPh _k , MVPv _k ) of the current PU, the available spatial and temporal correlation candidate motion vectors (MVCh _ki , MVCv _ki ), fitting coefficients ah _k , av _k and bh _k , bv _k dynamically adjust the basic search range SRBasic _k , and finally the final search range SR _k is jointly determined by the preset maximum search range SR _max , the minimum search range SR _min and the dynamically adjusted basic search range SRBasic _k . From the above description, it can be seen that since the motion estimation search range of the current PU is predicted by using the relevant candidate motion vectors in the space domain and the time domain, the search range can be dynamically adjusted in real time according to the actual situation, thereby reducing the amount of calculation, and also means reducing Search time.

The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: it can still be described in the foregoing embodiments Modifications are made to the recorded technical solutions, or equivalent replacements are made to some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims (10)

1. The motion estimation search range prediction method based on motion vector correlation, is characterized in that, comprises the following steps: S1. Presetting the maximum search range SR _max and the minimum search range SR _min ; S2. Calculate the fitting coefficients ah _k , av _k and bh _k , bv _k of the prediction unit PU in different modes, wherein the total number of modes of the prediction unit PU is Q, 0≤k<Q; S3. Obtain corresponding fitting coefficients ah _k , av _k and bh _k , bv _k according to the mode k of the current PU; S4. According to the mode k of the current PU, obtain the corresponding motion vector predictor (MVPh _k , MVPv _k ) and available spatial and temporal related candidate motion vectors (MVCh _ki , MVCv _ki ), where iCount _k represents the available spatial domain The total number of candidate motion vectors related to the time domain, 0≤i<iCount _k ; S5. Calculate the cumulative difference (ΔMVTh _k , ΔMVTv _k ) between the motion vector predictor (MVPh _k , MVPv _k ) and each available spatial and temporal domain related candidate motion vector (MVCh _ki , MVCv _ki ); S6. According to the fitting coefficients ah _k , av _k and bh _k , bv _k in S2, the motion vector predictor (MVPh _k , MVPv _k ) in S4, and the cumulative difference (ΔMVTh _k , ΔMVTv _k ) in S5, calculate basic search scope SRBasic _k ; S7. Determine the final search range SR _k according to the basic search range SRBasic _k in S6 and the maximum search range SR _max and minimum search range SR _min in S1. 2. The method according to claim 1, wherein said step S2 comprises: S21. Determine all modes of the prediction unit PU; S22. Obtain the available spatial and temporal motion-related candidate motion vectors (MVCh _ijk , MVCv _ijk ) of PU _jk in the test sequence, where 0≤k<Q, j represents the jth PU in the kth mode, and Q is The total number of modes of the prediction unit; S23. Obtain the motion vector predictor (MVPh _jk , MVPv _jk ) and the actual motion vector (MVh _jk , MVv _jk ) of the PU _jk ; S24. Calculate the difference between the motion vector predictor (MVPh _jk , MVPv _jk ) and each available space-domain and time-domain related candidate motion vector (MVCh _ijk , MVCv _ijk ), and sum after taking the absolute value: ${\mathrm{<span\; class="notranslate">\; \&Delta;MVTh</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; =</span>\underset{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 0</span>}}{\overset{{\mathrm{<span\; class="notranslate">\; iCount</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{<span\; class="notranslate">\; \&Sigma;</span>}}<span\; class="notranslate">\; |</span>{\mathrm{<span\; class="notranslate">\; MVCh</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}\mathrm{<span\; class="notranslate">\; j</span>}\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; MVPh</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; |</span>$ ${\mathrm{<span\; class="notranslate">\; \&Delta;MVTv</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; =</span>\underset{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 0</span>}}{\overset{{\mathrm{<span\; class="notranslate">\; iCount</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{<span\; class="notranslate">\; \&Sigma;</span>}}<span\; class="notranslate">\; |</span>{\mathrm{<span\; class="notranslate">\; MVC</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}\mathrm{<span\; class="notranslate">\; j</span>}\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; MVPv</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; |</span>$ Among them, iCount _jk represents the number of available spatial domain and temporal domain related candidate motion vectors; S25. Calculate the difference between the actual motion vector (MVh _jk , MVv _jk ) of PU _jk and the motion vector predictor (MVPh _jk , MVPv _jk ), and take the absolute value: ΔMVh _jk ＝|MVh _jk -MVPh _jk | ΔMVv _jk ＝|MVv _jk -MVPv _jk | S26, construct the following two equations: ah _k ×|MVPh _jk |+bh _k ×ΔMVTh _jk ＝ΔMVh _jk av _k ×|MVPv _jk |+bv _k ×ΔMVTv _jk ＝ΔMVv _jk ah _k , av _k and bh _k , bv _k are calculated by least squares fitting algorithm. 3. The method according to claim 1, wherein said step S5 comprises: Calculate the difference between the motion vector predictor (MVPh _k , MVPv _k ) and each available spatially and temporally correlated candidate motion vector (MVCh _ki , MVCv _ki ), taking the absolute value and summing: ${\mathrm{<span\; class="notranslate">\; \&Delta;MVTh</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; =</span>\underset{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 0</span>}}{\overset{{\mathrm{<span\; class="notranslate">\; iCount</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{<span\; class="notranslate">\; \&Sigma;</span>}}<span\; class="notranslate">\; |</span>{\mathrm{<span\; class="notranslate">\; MVCh</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; MVPh</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; |</span>$ ${\mathrm{<span\; class="notranslate">\; \&Delta;MVTv</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; =</span>\underset{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 0</span>}}{\overset{{\mathrm{<span\; class="notranslate">\; iCount</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{<span\; class="notranslate">\; \&Sigma;</span>}}<span\; class="notranslate">\; |</span>{\mathrm{<span\; class="notranslate">\; MVC</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; MVPv</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; |</span>$ Among them, iCount _k represents the number of available spatial and temporal correlation candidate motion vectors. 4. The method according to claim 1, wherein said step S6 comprises: Based on the fitting coefficients ah _k , av _k and bh _k , bv _k in S2, the motion vector predictor (MVPh _k , MVPv _k ) in S4, and the cumulative difference (ΔMVTh _k , ΔMVTv _k ) in S5, calculate the basic search scope: SRh _k-basic ＝|ah _k ×ΔMVTh _k +bh _k ×|MVPh _k || SRv _k-basic =|av _k ×ΔMVTv _k +bv _k ×|MVPv _k ||. SRBasic _k = SRh _k-basic + SRv _k-basic 5. The method according to claim 1, wherein said step S7 comprises: Determine the final search range SR _k according to the basic search range SRBasic _k and the preset maximum search range SR _max and minimum search range SR _min as follows: ${\mathrm{<span\; class="notranslate">\; SR</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; =</span>\left\{\begin{array}{cc}{\mathrm{<span\; class="notranslate">\; SR</span>}}_{\mathrm{<span\; class="notranslate">\; min</span>}}<span\; class="notranslate">\; ,</span>& {\mathrm{<span\; class="notranslate">\; SRBasic</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; <</span>{\mathrm{<span\; class="notranslate">\; SR</span>}}_{\mathrm{<span\; class="notranslate">\; min</span>}}\\ {\mathrm{<span\; class="notranslate">\; SRBasic</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; ,</span>& {\mathrm{<span\; class="notranslate">\; SR</span>}}_{\mathrm{<span\; class="notranslate">\; min</span>}}<span\; class="notranslate">\; \&le;</span>{\mathrm{<span\; class="notranslate">\; SRBasic</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; \&le;</span>{\mathrm{<span\; class="notranslate">\; SR</span>}}_{\mathrm{<span\; class="notranslate">\; max</span>}}\\ {\mathrm{<span\; class="notranslate">\; SR</span>}}_{\mathrm{<span\; class="notranslate">\; max</span>}}<span\; class="notranslate">\; ,</span>& {\mathrm{<span\; class="notranslate">\; SRBasic</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; ></span>{\mathrm{<span\; class="notranslate">\; SR</span>}}_{\mathrm{<span\; class="notranslate">\; max</span>}}\end{array}\right.<span\; class="notranslate">\; .</span>$ 6. A motion estimation search range prediction system based on motion vector correlation, characterized in that the system includes: The pre-search range setting module is used to pre-set the maximum search range SR _max and the minimum search range SR _min ; The fitting coefficient calculation module is used to calculate the fitting coefficients ah _k , av _k and bh _k , bv _k of the prediction unit PU in different modes, wherein the total number of modes of the prediction unit PU is Q, and 0≤k<Q; The fitting coefficient acquisition module of the corresponding mode is used to obtain the corresponding fitting coefficients ah _k , av _k and bh _k , bv _k according to the mode k of the current PU; The predictor and related motion vector acquisition module of the corresponding mode is used to acquire the corresponding motion vector predictor (MVPh _k , MVPv _k ) and available spatial and temporal related candidate motion vectors (MVCh _ki , MVCv _ki ), wherein, iCount _k represents the total number of available spatial domain and time domain related candidate motion vectors, 0≤i<iCount _k ; a correlation calculation module for calculating the cumulative difference (ΔMVTh _k , ΔMVTv _k ) between the motion vector predictor (MVPh _k , MVPv _k ) and each available spatial and temporal correlation candidate motion vector (MVCh _ki , MVCv _ki ); The basic search range calculation module is used to obtain the fitting coefficient ah _k , av _k and bh _k , bv _k in the module according to the fitting coefficient of the corresponding mode, the predictor of the corresponding mode and the motion vector prediction in the related motion vector acquisition module Sub (MVPh _k , MVPv _k ), the cumulative difference (ΔMVTh _k , ΔMVTv _k ) in the correlation calculation module, calculates the basic search range SRBasic _k ; The final search range determination module is used to determine the final search range SR _k according to the basic search range SRBasic _k in the basic search range calculation module and the maximum search range SR _max and the minimum search range SR _min in the pre-search range setting module. 7. system according to claim 6, is characterized in that, described fitting coefficient calculating module comprises: A PU mode determination unit, configured to determine all modes of the prediction unit PU; The related motion vector acquisition unit is used to acquire the available space domain and time domain motion related candidate motion vectors (MVCh _ijk , MVCv _ijk ) of PU _jk in the test sequence, 0≤k<Q, and j represents the kth mode in the kth mode j PUs, k is the mode of the current prediction unit; A predictor and actual motion vector acquisition unit, configured to acquire the motion vector predictor (MVPh _jk , MVPv _jk ) and actual motion vector (MVh _jk , MVv _jk ) of the PU _jk ; The correlation calculation unit is used to calculate the difference between the motion vector predictor (MVPh _jk , MVPv _jk ) and each available spatial and temporal correlation candidate motion vector (MVCh _ijk , MVCv _ijk ), and sum after taking the absolute value: ${\mathrm{<span\; class="notranslate">\; \&Delta;MVTh</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; =</span>\underset{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 0</span>}}{\overset{{\mathrm{<span\; class="notranslate">\; iCount</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{<span\; class="notranslate">\; \&Sigma;</span>}}<span\; class="notranslate">\; |</span>{\mathrm{<span\; class="notranslate">\; MVCh</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}\mathrm{<span\; class="notranslate">\; j</span>}\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; MVPh</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; |</span>$ ${\mathrm{<span\; class="notranslate">\; \&Delta;MVTv</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; =</span>\underset{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 0</span>}}{\overset{{\mathrm{<span\; class="notranslate">\; iCount</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{<span\; class="notranslate">\; \&Sigma;</span>}}<span\; class="notranslate">\; |</span>{\mathrm{<span\; class="notranslate">\; MVC</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}\mathrm{<span\; class="notranslate">\; j</span>}\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; MVPv</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; |</span>$ Among them, iCount _jk represents the number of available spatial domain and temporal domain related candidate motion vectors; The difference calculation unit is used to calculate the difference between the actual motion vector (MVh _jk , MVv _jk ) and the motion vector predictor (MVPh _jk , MVPv _jk ) of the PU _jk , and take the absolute value: ΔMVh _jk ＝|MVh _jk -MVPh _jk |; ΔMVv _jk ＝|MVv _jk -MVPv _jk | The coefficient fitting unit is used to construct the following two equation systems: ah _k ×|MVPh _jk |+bh _k ×ΔMVTh _jk ＝ΔMVh _jk av _k ×|MVPv _jk |+bv _k ×ΔMVTv _jk ＝ΔMVv _jk Ah _k , av _k and bh _k , bv _k are calculated by least squares fitting algorithm. 8. The system according to claim 6, wherein the correlation calculation module is used to calculate motion vector predictors (MVPh _k , MVPv _k ) and each available space and time domain correlation candidate motion vector (MVCh _ki , MVCv _ki ), take the absolute value and sum: ${\mathrm{<span\; class="notranslate">\; \&Delta;MVTh</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; =</span>\underset{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 0</span>}}{\overset{{\mathrm{<span\; class="notranslate">\; iCount</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{<span\; class="notranslate">\; \&Sigma;</span>}}<span\; class="notranslate">\; |</span>{\mathrm{<span\; class="notranslate">\; MVCh</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; MVPh</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; |</span>$ ${\mathrm{<span\; class="notranslate">\; \&Delta;MVTv</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; =</span>\underset{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 0</span>}}{\overset{{\mathrm{<span\; class="notranslate">\; iCount</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{<span\; class="notranslate">\; \&Sigma;</span>}}<span\; class="notranslate">\; |</span>{\mathrm{<span\; class="notranslate">\; MVC</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; MVPv</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; |</span>$ Among them, iCount _k represents the number of available spatial and temporal correlation candidate motion vectors. 9. system according to claim 6, is characterized in that, described basic search range calculation module is used for the fitting coefficient ah _k in the acquisition module according to the fitting coefficient of corresponding mode, av _k and bh _k , bv _k , The predictor of the corresponding mode and the motion vector predictor (MVPh _k , MVPv _k ) in the relevant motion vector acquisition module, the cumulative difference (ΔMVTh _k , ΔMVTv _k ) in the correlation calculation module, calculate the basic search range SRBasic _k : SRh _k-basic ＝|ah _k ×ΔMVTh _k +bh _k ×|MVPh _k || SRv _k-basic =|av _k ×ΔMVTv _k +bv _k ×|MVPv _k ||. SRBasic _k = SRh _k-basic + SRv _k-basic 10. The system according to claim 6, wherein the final search range determination module is used to determine the final search range according to the basic search range SRBasic _k and the preset maximum search range SR _max and minimum search range SR _min SR _k : ${\mathrm{<span\; class="notranslate">\; SR</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; =</span>\left\{\begin{array}{cc}{\mathrm{<span\; class="notranslate">\; SR</span>}}_{\mathrm{<span\; class="notranslate">\; min</span>}}<span\; class="notranslate">\; ,</span>& {\mathrm{<span\; class="notranslate">\; SRBasic</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; <</span>{\mathrm{<span\; class="notranslate">\; SR</span>}}_{\mathrm{<span\; class="notranslate">\; min</span>}}\\ {\mathrm{<span\; class="notranslate">\; SRBasic</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; ,</span>& {\mathrm{<span\; class="notranslate">\; SR</span>}}_{\mathrm{<span\; class="notranslate">\; min</span>}}<span\; class="notranslate">\; \&le;</span>{\mathrm{<span\; class="notranslate">\; SRBasic</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; \&le;</span>{\mathrm{<span\; class="notranslate">\; SR</span>}}_{\mathrm{<span\; class="notranslate">\; max</span>}}\\ {\mathrm{<span\; class="notranslate">\; SR</span>}}_{\mathrm{<span\; class="notranslate">\; max</span>}}<span\; class="notranslate">\; ,</span>& {\mathrm{<span\; class="notranslate">\; SRBasic</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; ></span>{\mathrm{<span\; class="notranslate">\; SR</span>}}_{\mathrm{<span\; class="notranslate">\; max</span>}}\end{array}\right.<span\; class="notranslate">\; .</span>$