CN112738517B - Motion estimation search method, device, equipment and storage medium - Google Patents
Motion estimation search method, device, equipment and storage medium Download PDFInfo
- Publication number
- CN112738517B CN112738517B CN201910974860.8A CN201910974860A CN112738517B CN 112738517 B CN112738517 B CN 112738517B CN 201910974860 A CN201910974860 A CN 201910974860A CN 112738517 B CN112738517 B CN 112738517B
- Authority
- CN
- China
- Prior art keywords
- macro block
- image
- current
- search
- initially selected
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/50—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using predictive coding
- H04N19/503—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using predictive coding involving temporal prediction
- H04N19/51—Motion estimation or motion compensation
- H04N19/56—Motion estimation with initialisation of the vector search, e.g. estimating a good candidate to initiate a search
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
- H04N19/102—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the element, parameter or selection affected or controlled by the adaptive coding
- H04N19/132—Sampling, masking or truncation of coding units, e.g. adaptive resampling, frame skipping, frame interpolation or high-frequency transform coefficient masking
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
- H04N19/169—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding
- H04N19/17—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding the unit being an image region, e.g. an object
- H04N19/176—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding the unit being an image region, e.g. an object the region being a block, e.g. a macroblock
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/50—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using predictive coding
- H04N19/503—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using predictive coding involving temporal prediction
- H04N19/51—Motion estimation or motion compensation
- H04N19/57—Motion estimation characterised by a search window with variable size or shape
Landscapes
- Engineering & Computer Science (AREA)
- Multimedia (AREA)
- Signal Processing (AREA)
- Compression Or Coding Systems Of Tv Signals (AREA)
Abstract
The invention discloses a motion estimation searching method, a device, equipment and a storage medium, wherein the method comprises the steps of respectively carrying out downsampling operation of corresponding multiples on a current image and a reference image to obtain a current downsampling image and a reference downsampling image; setting a corresponding search range in a reference sample image for pixel full search based on a current macro block of a current sample image to obtain a primary selection macro block with the minimum iteration cost with the current macro block; the macro block with the minimum iteration cost with the macro block in the current image is obtained by combining the initially selected macro block and the reference image, so that the large-range searching work is carried out in the down-sampling image by adopting the layered fast matching, and the reference image is only subjected to the small-range fine searching, thereby greatly reducing the calculation complexity and the calculation amount.
Description
Technical Field
The present invention relates to the field of motion estimation technology in video coding, and in particular, to a motion estimation search method, apparatus, device, and storage medium.
Background
H.264/AVC is a video compression standard proposed by the International telecommunication Union, video coding experts group and the ISO/IEC moving Picture experts group. Under the condition of reconstructing the same image quality, the coding efficiency of H.264 is improved by 48.8 percent and 38.62 percent respectively compared with that of H.263 and MPEG-4. However, the improvement of the h.264 coding efficiency comes at the cost of an increase in computational complexity. The search for a more effective quick search algorithm is of great significance to reduce the complexity of operation and improve the performance of the algorithm.
In video compression, Motion Estimation (ME) is one of the key technologies for improving compression efficiency, and aims to extract information about object Motion from video images, and is the most basic and effective method for eliminating redundancy. Among motion estimation algorithms, the simplest algorithm is the Full Search (FS), which is to compare all possible macroblocks in a Search area and select the minimum matching error value as a predicted value, and the corresponding offset is the motion vector to be obtained. Because the algorithm search has ergodicity, the calculation amount is inevitably too large, and occupies more than 80% of the whole H.264 calculation amount, so that the calculation of motion estimation is complex, and the coding compression efficiency is low.
Disclosure of Invention
It is a primary object of the present invention to provide a motion estimation search method, apparatus, device and storage medium, which overcome the above technical problems.
According to a first aspect of the present invention, there is provided a motion estimation search method, the method comprising: respectively carrying out downsampling operation of corresponding multiples on the current image and the reference image to obtain a current downsampling image and a reference downsampling image; setting a corresponding search range in the reference sample image and performing pixel full search based on the current macro block of the current sample image to obtain a primary selection macro block with the minimum iteration cost with the current macro block; and combining the initially selected macro block and the reference image to obtain the macro block with the minimum iteration cost with the macro block in the current image.
Optionally, the performing down-sampling operations of corresponding multiples on the current image and the reference image respectively includes: 1/2 downsampling operation is performed on the current image and the reference image in two dimensions.
Optionally, the setting, based on the current macro block of the current sample image, a corresponding search range in the reference sample image and performing pixel full search to obtain an initially selected macro block with a minimum iteration cost with the current macro block includes: establishing a first search frame with a corresponding size on the reference sampling image based on a current macro block of the current sampling image, wherein the current macro block of the current sampling image is the center of the first search frame; and performing integer pixel full search in the first search frame to find the initially selected macro block with the minimum sum of the bit number of the coded MVD of the current macro block in the current sampling image and the sum absolute error SAD.
Optionally, the obtaining, by combining the initially selected macroblock and the reference image, a macroblock with a minimum macroblock iteration cost in the current image includes: and performing pixel full search in the reference image based on the initially selected macro block to obtain a preferred macro block with the minimum sum of the bit number of the coded MVD of the initially selected macro block and the sum absolute error SAD, and taking the preferred macro block as the macro block with the minimum iteration cost with the macro block in the current image.
Optionally, the performing a full pixel search in the reference image to obtain a preferred macroblock with a minimum sum of the sum absolute error SAD and the number of bits of the coded MVD of the initially selected macroblock, includes: establishing a second search box with a corresponding size on the reference image based on the initially selected macro block, wherein the initially selected macro block is the center of the second search box; and performing integer pixel full search in the second search frame to obtain a preferred macro block with the minimum sum of the bit number of the coded MVD of the initially selected macro block and the sum absolute error SAD.
Optionally, the obtaining, by combining the initially selected macroblock and the reference image, a macroblock with a minimum macroblock iteration cost in the current image includes: performing pixel full search in the reference image based on the initially selected macro block to obtain a preferred macro block with the minimum sum of the bit number of the coded MVD of the initially selected macro block and the sum absolute error SAD; in the horizontal direction, performing pixel interpolation processing of a set multiple on the reference image to obtain an interpolation processing image; establishing a third search frame with a corresponding size on the interpolation processing image based on the preferred macro block, wherein the preferred macro block is the center of the third search frame; and performing corresponding multiple pixel full search in the third search frame to obtain an optimal macro block with the minimum sum of the bit number of the coded MVD of the optimal macro block and the sum absolute error SAD, and taking the optimal macro block as the macro block with the minimum iteration cost with the macro block in the current image.
According to a second aspect of the present invention, there is provided a motion estimation fast search apparatus, comprising: the down-sampling module is used for respectively performing down-sampling operation of corresponding multiples on the current image and the reference image to obtain a current sample unloading image and a reference sample unloading image; the initial selection module is used for setting a corresponding search range in the reference sample image and carrying out pixel full search based on the current macro block of the current sample image to obtain an initial selection macro block with the minimum iteration cost with the current macro block; and the fine selection module is used for combining the initially selected macro block and the reference image to obtain the macro block with the minimum iteration cost with the macro block in the current image.
Optionally, the preliminary selection module includes: a first search frame establishing unit, configured to establish a first search frame with a corresponding size on the reference downsampled image based on a current macro block of the current downsampled image, where the current macro block of the current downsampled image is a center of the first search frame; and the first search unit is used for performing integer pixel full search in the first search frame so as to find the initially selected macro block with the minimum sum of the bit number of the coded MVD of the current macro block in the current sample image and the sum absolute error SAD.
According to a third aspect of the present invention, there is provided a computer device comprising a processor and a memory; the memory is used for storing computer instructions, and the processor is used for executing the computer instructions stored by the memory to realize the motion estimation searching method.
According to a fourth aspect of the present invention, there is provided a computer readable storage medium storing one or more programs, the one or more programs being executable by one or more processors to implement a motion estimation search method as described above.
The invention has the following beneficial effects: by adopting layered fast matching, large-range searching work is carried out in the down-sampling image, and the reference image is only subjected to small-range fine searching, so that the calculation complexity is greatly reduced and the calculation amount is reduced.
Drawings
FIG. 1 is a block diagram of a motion estimation search method according to the present invention;
FIG. 2 is a simplified diagram of a first search box according to the present invention;
FIG. 3 is a schematic diagram of pixel interpolation processing in the present invention;
FIG. 4 is a flowchart illustrating an exemplary motion estimation search method according to the present invention;
fig. 5 is a schematic structural diagram of the motion estimation fast search apparatus according to the present invention.
The implementation, functional features and advantages of the objects of the present invention will be further explained with reference to the accompanying drawings.
Detailed Description
It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
In the following description, suffixes such as "module", "component", or "unit" used to denote elements are used only for facilitating the explanation of the present invention, and have no specific meaning in itself. Thus, "module", "component" or "unit" may be used mixedly.
In order to facilitate understanding of the embodiments of the present invention, the following detailed description of the embodiments of the present invention is provided.
A first embodiment of the present invention provides a motion estimation search method, including: respectively carrying out downsampling operation of corresponding multiples on the current image and the reference image to obtain a current downsampling image and a reference downsampling image; setting a corresponding search range in the reference sample image and performing pixel full search based on the current macro block of the current sample image to obtain a primary selection macro block with the minimum iteration cost with the current macro block; and combining the initially selected macro block and the reference image to obtain the macro block with the minimum iteration cost with the macro block in the current image.
In this regard, by using hierarchical fast matching, large-scale search work is performed in the downsampled image, while the reference image is only subjected to small-scale fine search, which greatly reduces the computational complexity and the amount of computation.
Fig. 1 is a flow chart of a motion estimation search method according to the present invention. Specifically, as shown in fig. 1, a first embodiment of the present invention provides a motion estimation search method, where the method includes:
s11: respectively carrying out downsampling operation of corresponding multiples on the current image and the reference image to obtain a current downsampling image and a reference downsampling image;
in this embodiment, the direction and size of the downsampling operation are not limited, and only the requirements of this embodiment need to be satisfied.
In another embodiment, one implementation of this step S11 is: 1/2 downsampling operation is performed on the current image and the reference image in two dimensions. Specifically, 1/2 downsampling operations are performed on the current image and the reference image in the x/y direction (including the horizontal direction and the vertical direction), respectively. Namely: the width and height of the current image and the width and height of the reference image are reduced to half, respectively.
S12: setting a corresponding search range in the reference sample image and performing pixel full search based on the current macro block of the current sample image to obtain a primary selection macro block with the minimum iteration cost with the current macro block;
in the present embodiment, a search range is set in the reference sample image, so that the motion vector and the summed absolute error SAD of the position of the current macroblock in the current sample image in the reference sample image are found in the search range.
Then, a primary selected macro block which is the sum of the bit number of the coded MVD of the current macro block and the sum absolute error SAD is selected in a selected area in the reference sample image. Thereby realizing the pre-fetching work of the macro block.
Fig. 2 is a simple schematic diagram of a first search box in the present invention, specifically, according to fig. 2, in another embodiment, an implementation manner of S12 includes:
s121: establishing a first search frame with a corresponding size on the reference sampling image based on a current macro block of the current sampling image, wherein the current macro block of the current sampling image is the center of the first search frame;
s122: and performing integer pixel full search in the first search frame to find the initially selected macro block with the minimum sum of the bit number of the coded MVD of the current macro block in the current sampling image and the sum absolute error SAD.
Specifically, a search range is set in the reference downsampled image by centering on a current macro block of the current downsampled image, such as: a first search box having a search range of y e-12, 11, x e-16, 15, is set in a reference downsampled image centering on a current macro block of the current downsampled image. In this embodiment, the size of the search range is not limited, and only needs to satisfy the requirements of this embodiment.
Thus, the motion vector and the summed absolute error SAD with the position of the current macroblock in the current downsampled image in the reference downsampled image are found in the search range.
And then, performing integer pixel full search in a first search frame to find the initially selected macro block with the minimum sum of the bit number of the coded MVD of the current macro block in the current sample image and the sum absolute error SAD. Therefore, the initially selected macro block with the minimum iteration cost with the current macro block in the current sampling image is obtained.
S13: and combining the initially selected macro block and the reference image to obtain the macro block with the minimum iteration cost with the macro block in the current image.
In this embodiment, a macroblock with the minimum iteration cost from the macroblock in the current picture can be obtained by combining the initially selected macroblock with the reference picture, that is: after the pre-selection work is carried out in the reference sample image, the fine search is carried out on the reference image, which is beneficial to reducing the complexity of calculation.
In this regard, by using hierarchical fast matching, large-scale search work is performed in the downsampled image, while the reference image is only subjected to small-scale fine search, which greatly reduces the computational complexity and the amount of computation.
In another embodiment, one implementation manner of the step S13 includes:
and performing pixel full search in the reference image based on the initially selected macro block to obtain a preferred macro block with the minimum sum of the bit number of the coded MVD of the initially selected macro block and the sum absolute error SAD, and taking the preferred macro block as the macro block with the minimum iteration cost with the macro block in the current image.
Specifically, the step S13 includes:
s131: establishing a second search box with a corresponding size on the reference image based on the initially selected macro block, wherein the initially selected macro block is the center of the second search box;
s132: and performing integer pixel full search in the second search frame to obtain a preferred macro block with the minimum sum of the bit number of the coded MVD of the initially selected macro block and the sum absolute error SAD.
Specifically, a search range is set by taking the initially selected macro block as a center in the reference image, such as: and setting a second search box with a search range of y ∈ [ -2,2], x ∈ [ -2,2] by taking the initially selected macro block as the center in the reference image. In this embodiment, the size of the search range is not limited, and only needs to satisfy the requirements of this embodiment.
Thus, the motion vector and the sum absolute error SAD of the position of the initially selected macro block in the reference image are found in the search range.
Then, a full-pel search is performed in a second search box to find a preferred macroblock with the minimum sum of the sum absolute error SAD and the number of bits of coded MVDs of the initially selected macroblock. Therefore, a preferred macro block with the minimum iteration cost with the initially selected macro block is obtained, and the preferred macro block is used as the macro block with the minimum iteration cost with the macro block in the current image.
In this regard, by using hierarchical fast matching, large-scale search work is performed in the downsampled image, while the reference image is only subjected to small-scale fine search, which greatly reduces the computational complexity and the amount of computation. And moreover, full search is adopted in the down-sampling image and the reference image, so that the accuracy of the motion vector is ensured.
Specifically, in another embodiment, another implementation manner of the step S13 includes:
s133: performing pixel full search in the reference image based on the initially selected macro block to obtain a preferred macro block with the minimum sum of the bit number of the coded MVD of the initially selected macro block and the sum absolute error SAD;
s134: in the horizontal direction, performing pixel interpolation processing of a set multiple on the reference image to obtain an interpolation processing image;
fig. 3 is a schematic diagram of the pixel interpolation process in the present invention. Specifically, as shown in fig. 3, 1/4 pixel interpolation is performed on the reference image in the x direction, but in this embodiment, the preset multiple value is not limited, and only needs to meet the requirement of this embodiment. As shown in fig. 2, the pixels a, b, and c are interpolated by referring to the peripheral pixel E, F, G, H, I, J.
Of course, in this embodiment, the step S134 only needs to be performed before the following step S136, and the execution timing thereof is not limited, such as: this step S134 may be performed at the above-described step S11.
S135: establishing a third search frame with a corresponding size on the interpolation processing image based on the preferred macro block, wherein the preferred macro block is the center of the third search frame;
s136: and performing corresponding multiple pixel full search in the third search frame to obtain an optimal macro block with the minimum sum of the bit number of the coded MVD of the optimal macro block and the sum absolute error SAD, and taking the optimal macro block as the macro block with the minimum iteration cost with the macro block in the current image.
Specifically, a search range is set by centering on a preferred macro block in an interpolation-processed image, such as: a third search box having a search range of y ∈ [0,0], x ∈ [ 0.5,0.5] is set in the interpolation-processed image centering on the preferred macro block. In this embodiment, the size of the search range is not limited, and only needs to satisfy the requirements of this embodiment.
Thus, the motion vector and the summed absolute error SAD with the position of the preferred macroblock in the interpolated image are found within the search range.
Then, a full-pel search is performed in a third search box to find an optimal macroblock with the minimum sum of the number of bits of the coded MVD and the sum absolute error SAD of the optimal macroblock. Therefore, the optimal macro block with the minimum iteration cost with the optimal macro block is obtained, and the optimal macro block is used as the macro block with the minimum iteration cost with the macro block in the current image.
In this regard, through the above-described steps S121 to S135, the search accuracy is gradually improved based on the hierarchical search, and moreover, it is achieved that the current macro block is compared with all candidate blocks within the search window to find the best matching macro block.
Fig. 4 is a flowchart illustrating an exemplary motion estimation search method according to the present invention. Specifically, as shown in fig. 4, in order to facilitate understanding of the technical solution of the present invention, the following embodiments are exemplarily described:
s1, carrying out 1/2 down-sampling on the current image orig _ pic in the x/y direction to obtain a current sample image scale _ orig _ pic;
s2, down-sampling the reference image ref _ pic in the x/y direction 1/2 to obtain a reference sample image scale _ ref _ pic;
s3, performing 1/4 pixel interpolation on ref _ pic in the x direction to obtain an interpolation processing image sub _ ref _ pic;
s4, on scale _ ref _ pic, using the position of the current macro block on scale _ orig _ pic as the center of the search window, performing full search of integer pixel in the search frame, and finding out the initially selected macro block with the minimum iteration cost with the current macro block;
specifically, the search window is expanded up, down, left, and right along the X/Y axis with the position of the current macro block as the center.
Wherein the search window size is y ∈ [ -12,11], x ∈ [ -16,15 ].
Also, the block matching criterion in this step S4 is the summed absolute error SAD + MVD _ BIN. Specifically, the sum Absolute error sad (sum of Absolute difference), which is the sum Absolute error, is the sum of the Absolute values of the residuals of each pixel (i.e., the pixel value difference between the current macroblock and the candidate block) and is equal to Σ abs (org _ val-ref _ val); MVD _ BIN is the number of bits that encodes MVD (motion Vectors difference). Finding out the initial selection macro block with the minimum iteration cost (namely the minimum sum absolute error SAD + MVD _ BIN value) by comparing the current macro block with all candidate blocks in the search box;
s5, taking the initially selected macro block obtained in S4 as the center, establishing a corresponding search box on ref _ pic, and performing full search on the whole pixel point in the search box to find out the optimal macro block with the minimum iteration cost with the initially selected macro block;
the search window is expanded up, down, left and right along the X/Y axis with the position of the primary selection module as the center.
Wherein, the establishment carries out the whole pixel point full search according to the minimum criterion of the sum absolute error SAD + MVD _ BIN (namely the minimum value of the sum absolute error SAD + MVD _ BIN), finds out the preferred macro block in the search range, and the size of the search window is y ∈ [ -2,2], x ∈ [ -2,2]
S6, establishing a corresponding search box on sub _ ref _ pic by taking the preferred macro block obtained in S5 as a center, performing full search on the whole pixel point in the search box, and finding out the optimal macro block with the minimum iteration cost of the preferred macro block;
the search window is centered on the position of the preferred macroblock and extends left and right along the X-axis.
Wherein the search window size is y ∈ [0,0], x ∈ [ 0.5,0.5 ]. And (3) carrying out 1/4 pixel point full search on the sub _ ref _ pic according to the minimum criterion of the sum absolute error SAD + MVD _ BIN (namely the minimum value of the sum absolute error SAD + MVD _ BIN), and finding out the optimal macro block in the search range.
Fig. 5 is a schematic structural diagram of the motion estimation fast search apparatus according to the present invention. According to fig. 5, a second embodiment of the present invention provides a motion estimation fast search apparatus, including: the down-sampling module 110 is configured to perform down-sampling operations on the current image and the reference image by corresponding multiples respectively to obtain a current down-sampling image and a reference down-sampling image; a primary selection module 111, configured to set a corresponding search range in the reference sample image based on a current macro block of the current sample image, and perform pixel full search to obtain a primary selection macro block with a minimum iteration cost with the current macro block; and a refinement module 112, configured to combine the initially selected macroblock and the reference image to obtain a macroblock with a minimum iteration cost from the macroblock in the current image.
Optionally, the down-sampling module 110 is specifically configured to: 1/2 downsampling operation is performed on the current image and the reference image in two dimensions.
Optionally, the primary selection module 111 includes: a first search frame establishing unit, configured to establish a first search frame with a corresponding size on the reference downsampled image based on a current macro block of the current downsampled image, where the current macro block of the current downsampled image is a center of the first search frame; and the first search unit is used for performing integer pixel full search in the first search frame so as to find the initially selected macro block with the minimum sum of the bit number of the coded MVD of the current macro block in the current sample image and the sum absolute error SAD.
Optionally, the culling module 112 includes: and the reference image searching unit is used for carrying out pixel full search in the reference image based on the initially selected macro block to obtain a preferred macro block with the minimum sum of the bit number of the coded MVD of the initially selected macro block and the sum absolute error SAD, and taking the preferred macro block as the macro block with the minimum iteration cost with the macro block in the current image.
Optionally, the reference image searching unit is specifically configured to: establishing a second search box with a corresponding size on the reference image based on the initially selected macro block, wherein the initially selected macro block is the center of the second search box; and performing integer pixel full search in the second search frame to obtain a preferred macro block with the minimum sum of the bit number of the coded MVD of the initially selected macro block and the sum absolute error SAD.
Optionally, the culling module 112 includes: the optimization unit is used for carrying out pixel full search in the reference image based on the initially selected macro block to obtain an optimized macro block with the minimum sum of the bit number of the coded MVD of the initially selected macro block and the sum absolute error SAD; the interpolation processing module is used for performing pixel interpolation processing of a set multiple on the reference image in the horizontal direction after the optimal macro block is obtained to obtain an interpolation processing image; an interpolation image search frame establishing module, configured to establish a third search frame with a corresponding size on the interpolation processing image based on the preferred macro block, where the preferred macro block is a center of the third search frame; and the interpolation image searching module is used for carrying out pixel full search of corresponding multiples in the third search frame to obtain an optimal macro block with the minimum sum of the bit number of the coded MVD of the optimal macro block and the sum absolute error SAD, and taking the optimal macro block as the macro block with the minimum iteration cost with the macro block in the current image.
A third embodiment of the present invention provides a computer device comprising a processor and a memory; the memory is used for storing computer instructions, and the processor is used for executing the computer instructions stored by the memory to realize the motion estimation searching method.
The terms and implementation principles related to the electronic device in the third embodiment of the present invention may specifically refer to a motion estimation search method in the first embodiment of the present invention, and are not described herein again.
A fourth embodiment of the present invention provides a computer-readable storage medium storing one or more modules executable by one or more processors to implement a motion estimation search method as described above.
The terms and implementation principles related to a computer-readable storage medium in the fourth embodiment of the present invention may specifically refer to a motion estimation search method in the first embodiment of the present invention, and are not described herein again.
It should be noted that, in this document, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other like elements in a process, method, article, or apparatus that comprises the element.
The above-mentioned serial numbers of the embodiments of the present invention are merely for description and do not represent the merits of the embodiments.
Through the above description of the embodiments, those skilled in the art will clearly understand that the method of the above embodiments can be implemented by software plus a necessary general hardware platform, and certainly can also be implemented by hardware, but in many cases, the former is a better implementation manner. Based on such understanding, the technical solutions of the present invention may be embodied in the form of a software product, which is stored in a storage medium (such as ROM/RAM, magnetic disk, optical disk) and includes instructions for enabling a terminal (such as a mobile phone, a computer, a server, an air conditioner, or a network device) to execute the method according to the embodiments of the present invention.
While the present invention has been described with reference to the embodiments shown in the drawings, the present invention is not limited to the embodiments, which are illustrative and not restrictive, and it will be apparent to those skilled in the art that various changes and modifications can be made therein without departing from the spirit and scope of the invention as defined in the appended claims.
Claims (8)
1. A method for motion estimation search, the method comprising:
respectively carrying out downsampling operation of corresponding multiples on the current image and the reference image to obtain a current downsampling image and a reference downsampling image;
setting a corresponding search range in the reference sample image and performing pixel full search based on the current macro block of the current sample image to obtain a primary selected macro block with the minimum iteration cost of the current macro block, and setting a corresponding search range in the reference sample image and performing pixel full search based on the current macro block of the current sample image to obtain a primary selected macro block with the minimum iteration cost of the current macro block, including: establishing a first search frame with a corresponding size on the reference sampling image based on a current macro block of the current sampling image, wherein the current macro block of the current sampling image is the center of the first search frame; performing full-pixel search in the first search frame to find a primary selected macro block with the minimum sum of the bit number of the coded MVD of the current macro block in the current sampling image and the sum absolute error SAD;
and combining the initially selected macro block and the reference image to obtain the macro block with the minimum iteration cost with the macro block in the current image.
2. The method according to claim 1, wherein the down-sampling operation on the current picture and the reference picture by corresponding multiples respectively comprises:
1/2 downsampling operation is performed on the current image and the reference image in two dimensions.
3. The method according to any one of claims 1-2, wherein said combining the initially selected macroblock and the reference picture to obtain the macroblock with the minimum iteration cost with the macroblock in the current picture comprises:
and performing pixel full search in the reference image based on the initially selected macro block to obtain a preferred macro block with the minimum sum of the bit number of the coded MVD of the initially selected macro block and the sum absolute error SAD, and taking the preferred macro block as the macro block with the minimum iteration cost with the macro block in the current image.
4. The method of claim 3, wherein the performing a full pixel search in the reference picture to obtain a preferred macroblock with a minimum sum of the sum absolute difference SAD and the number of bits of the coded MVD of the initially selected macroblock comprises:
establishing a second search box with a corresponding size on the reference image based on the initially selected macro block, wherein the initially selected macro block is the center of the second search box;
and performing integer pixel full search in the second search frame to obtain a preferred macro block with the minimum sum of the bit number of the coded MVD of the initially selected macro block and the sum absolute error SAD.
5. The method according to any one of claims 1-2, wherein said combining the initially selected macroblock and the reference picture to obtain the macroblock with the minimum iteration cost with the macroblock in the current picture comprises:
performing pixel full search in the reference image based on the initially selected macro block to obtain a preferred macro block with the minimum sum of the bit number of the coded MVD of the initially selected macro block and the sum absolute error SAD;
in the horizontal direction, performing pixel interpolation processing of a set multiple on the reference image to obtain an interpolation processing image;
establishing a third search frame with a corresponding size on the interpolation processing image based on the preferred macro block, wherein the preferred macro block is the center of the third search frame;
and performing pixel full search of corresponding multiples in the third search frame to obtain an optimal macro block with the minimum sum of the bit number of the coded MVD of the optimal macro block and the sum absolute error SAD, and taking the optimal macro block as the macro block with the minimum iteration cost with the macro block in the current image.
6. An apparatus for fast search of motion estimation, the apparatus comprising:
the down-sampling module is used for respectively performing down-sampling operation of corresponding multiples on the current image and the reference image to obtain a current sample unloading image and a reference sample unloading image;
a primary selection module, configured to set a corresponding search range in the reference sample image based on a current macro block of the current sample image, and perform pixel full search to obtain a primary selection macro block with a minimum iteration cost with the current macro block, where the primary selection module includes: a first search frame establishing unit, configured to establish a first search frame with a corresponding size on the reference downsampled image based on a current macro block of the current downsampled image, where the current macro block of the current downsampled image is a center of the first search frame; the first search unit is used for performing integer pixel full search in the first search frame so as to find the initially selected macro block with the minimum sum of the bit number of the coded MVD of the current macro block in the current sample image and the sum absolute error SAD;
and the fine selection module is used for combining the initially selected macro block and the reference image to obtain the macro block with the minimum iteration cost with the macro block in the current image.
7. A computer device comprising a processor and a memory;
the memory is configured to store computer instructions and the processor is configured to execute the computer instructions stored by the memory to implement the motion estimation search method of any of claims 1 to 5.
8. A computer-readable storage medium storing one or more programs, the one or more programs being executable by one or more processors to implement the motion estimation search method of any one of claims 1 to 5.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910974860.8A CN112738517B (en) | 2019-10-14 | 2019-10-14 | Motion estimation search method, device, equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910974860.8A CN112738517B (en) | 2019-10-14 | 2019-10-14 | Motion estimation search method, device, equipment and storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112738517A CN112738517A (en) | 2021-04-30 |
| CN112738517B true CN112738517B (en) | 2022-03-01 |
Family
ID=75588572
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910974860.8A Active CN112738517B (en) | 2019-10-14 | 2019-10-14 | Motion estimation search method, device, equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112738517B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114531596A (en) * | 2022-01-25 | 2022-05-24 | 京东方科技集团股份有限公司 | Image processing method and device |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101605262A (en) * | 2009-07-09 | 2009-12-16 | 杭州士兰微电子股份有限公司 | The predicting size motion of variable block method and apparatus |
| CN103139562A (en) * | 2011-11-30 | 2013-06-05 | 富士通株式会社 | Motion estimation method and motion estimation device |
| CN105007493A (en) * | 2015-07-17 | 2015-10-28 | 珠海市杰理科技有限公司 | Match searching method and system based on motion search window |
| CN108900846A (en) * | 2018-07-17 | 2018-11-27 | 珠海亿智电子科技有限公司 | A kind of the two-dimensional directional motion estimation hardware circuit and its method of Video coding |
| WO2019072370A1 (en) * | 2017-10-09 | 2019-04-18 | Huawei Technologies Co., Ltd. | Memory access window and padding for motion vector refinement |
-
2019
- 2019-10-14 CN CN201910974860.8A patent/CN112738517B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101605262A (en) * | 2009-07-09 | 2009-12-16 | 杭州士兰微电子股份有限公司 | The predicting size motion of variable block method and apparatus |
| CN103139562A (en) * | 2011-11-30 | 2013-06-05 | 富士通株式会社 | Motion estimation method and motion estimation device |
| CN105007493A (en) * | 2015-07-17 | 2015-10-28 | 珠海市杰理科技有限公司 | Match searching method and system based on motion search window |
| WO2019072370A1 (en) * | 2017-10-09 | 2019-04-18 | Huawei Technologies Co., Ltd. | Memory access window and padding for motion vector refinement |
| CN108900846A (en) * | 2018-07-17 | 2018-11-27 | 珠海亿智电子科技有限公司 | A kind of the two-dimensional directional motion estimation hardware circuit and its method of Video coding |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112738517A (en) | 2021-04-30 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11089333B2 (en) | Method for deriving a temporal predictive motion vector, and apparatus using the method | |
| US11082713B2 (en) | Method and apparatus for global motion compensation in video coding system | |
| US8705611B2 (en) | Image prediction encoding device, image prediction encoding method, image prediction encoding program, image prediction decoding device, image prediction decoding method, and image prediction decoding program | |
| EP1430724B1 (en) | Motion estimation and/or compensation | |
| US8503532B2 (en) | Method and apparatus for inter prediction encoding/decoding an image using sub-pixel motion estimation | |
| JP3538055B2 (en) | Motion vector detection device | |
| US20060203912A1 (en) | Motion vector detection method, motion vector detection apparatus, computer program for executing motion vector detection process on computer | |
| CN102804779A (en) | Image processing device and method | |
| EP1960967A1 (en) | Motion estimation using prediction guided decimated search | |
| US8532409B2 (en) | Adaptive motion search range determining apparatus and method | |
| US7505636B2 (en) | System and method for two-pass interpolation for quarter-pel motion compensation | |
| EP2465254A1 (en) | Techniques to perform video stabilization and detect video shot boundaries based on common processing elements | |
| CN112738517B (en) | Motion estimation search method, device, equipment and storage medium | |
| CN116368799A (en) | Enhanced motion estimation for inter-frame coding | |
| EP2953362B1 (en) | Moving image coding device | |
| CN114040209A (en) | Motion estimation method, motion estimation device, electronic equipment and storage medium | |
| EP1683361B1 (en) | Power optimized collocated motion estimation method | |
| AU2015200360B2 (en) | Method for deriving a temporal predictive motion vector, and apparatus using the method | |
| WO2023020591A1 (en) | Method and apparatus for hardware-friendly template matching in video coding system | |
| WO2023020390A1 (en) | Method and apparatus for low-latency template matching in video coding system | |
| CN117939147B (en) | Video encoding and decoding device | |
| EP4099698A1 (en) | Method, apparatus and device for constructing motion information list in video coding and decoding | |
| Chiang et al. | Fast multi-reference frame motion estimation via downhill simplex search | |
| JPH07288817A (en) | Motion vector detector | |
| JP3458473B2 (en) | Apparatus and method for detecting motion vector |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |