CN112714316B - Regular mark detection and classification identification method based on video code stream - Google Patents
Regular mark detection and classification identification method based on video code stream Download PDFInfo
- Publication number
- CN112714316B CN112714316B CN202011522015.6A CN202011522015A CN112714316B CN 112714316 B CN112714316 B CN 112714316B CN 202011522015 A CN202011522015 A CN 202011522015A CN 112714316 B CN112714316 B CN 112714316B
- Authority
- CN
- China
- Prior art keywords
- value
- mark
- intra
- rule
- frame
- 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
- H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
- H04N19/134—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the element, parameter or criterion affecting or controlling the adaptive coding
- H04N19/157—Assigned coding mode, i.e. the coding mode being predefined or preselected to be further used for selection of another element or parameter
- H04N19/159—Prediction type, e.g. intra-frame, inter-frame or bidirectional frame prediction
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/40—Scenes; Scene-specific elements in video content
- G06V20/41—Higher-level, semantic clustering, classification or understanding of video scenes, e.g. detection, labelling or Markovian modelling of sport events or news items
-
- 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
- H04N19/13—Adaptive entropy coding, e.g. adaptive variable length coding [AVLC] or context adaptive binary arithmetic coding [CABAC]
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Multimedia (AREA)
- Signal Processing (AREA)
- Computational Linguistics (AREA)
- Software Systems (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Compression Or Coding Systems Of Tv Signals (AREA)
Abstract
A rule mark detection and classification identification method based on video code streams. The invention relates to the technical field of mark detection and identification. A method for realizing rule mark detection and classification identification is characterized in that the rule marks are detected and classified and identified through an obtained video analysis video code stream intra-frame prediction coding code stream prediction mode, a pixel residual error distribution condition value, a pixel residual error value, an inter-frame prediction code stream prediction mode, a motion vector and a pixel residual error distribution condition value. The invention avoids the problem of low detection precision due to illumination change and improves the detection precision; the method avoids time-consuming operations such as integer IDCT transformation, inverse quantization, reconstruction, loop filtering and the like in the video decoding process, and is beneficial to real-time detection of the rule marks.
Description
Technical Field
The invention relates to the technical field of mark detection and identification.
Background
The regular marks are marks which have regular geometric shapes and are obviously different from the surrounding environment and play roles of warning, reminding, indicating and the like, such as various regular marks in the traffic field, various marks of hazard sources and the like. The mark detection has wide application fields in actual life, and the effects of mark detection and identification are required to be guaranteed in intelligent traffic, intelligent video monitoring, object tracking and the like. For example, on the background that traffic jam and accidents are increasingly frequent, automatic detection and identification of the rule marks are important components of an intelligent traffic system, the rule marks can be accurately identified in time and drivers can be reminded, traffic accidents are avoided, and the method has great practical significance in the aspect of traffic safety operation. The conventional rule mark detection and identification method mainly comprises a traditional image detection and identification method and a rule mark detection and identification method based on deep learning. The traditional image detection and identification method is easily influenced by the environment, has poor adaptability to the environment and cannot meet the requirement of high-precision real-time identification; the rule mark detection and identification method based on deep learning has high identification precision, and in order to achieve the high identification precision, a neural network adopted by the deep learning has a complex structure, numerous network parameters and high calculation complexity, and is difficult to meet the real-time requirement of a system. The processing objects of the two methods are still images, the actual monitoring data are data after video compression, and the video data need to be completely decoded and then further processed; the two methods only consider the spatial characteristics of the regular marks when the processing object is a still image, and the actual video sequence contains a plurality of the temporal characteristics of the regular marks and cannot be well utilized.
Disclosure of Invention
The technical problem to be solved by the invention is as follows: how to directly utilize code stream characteristic information in the video code stream, the conventional method is not required to be adopted for detecting and identifying the rule marks after the video code stream is completely decoded, and time-consuming operations such as inverse quantization, integer IDCT conversion, reconstruction, loop filtering and the like in the decoding process are avoided.
The technical scheme adopted by the invention is as follows: a method for detecting and classifying and identifying rule marks based on video code streams comprises the steps of obtaining an intra-frame prediction mode, a pixel residual error distribution condition CBP value and a pixel residual error value of an intra-frame prediction coding code stream of a video code stream and an inter-frame prediction mode, a motion vector residual error value and an inter-frame pixel residual error distribution condition CBP value of an inter-frame prediction coding code stream of the video code stream through obtained video analysis, detecting and classifying and identifying the rule marks, and concretely carrying out the following steps
Dividing each frame of image of the obtained video into a plurality of blocks according to 4x4 pixels, wherein each block is defined as a basic block, partially decoding an intra-frame prediction coding code stream of a video code stream, wherein the partial decoding process comprises entropy decoding and inverse scanning of the intra-frame prediction code stream, and does not perform inverse quantization, integer IDCT transformation, reconstruction and loop filtering to obtain a macro block type, a quantization parameter, an intra-frame prediction mode, a CBP value of intra-frame pixel residual error distribution condition and a pixel residual error value of the basic block;
step two, judging whether the basic block is a regular mark edge part or not according to the intra-frame prediction mode of the basic block, the intra-frame pixel residual error distribution condition CBP value and the pixel residual error value information, if the intra-frame prediction mode of the basic block and the intra-frame pixel residual error distribution condition CBP value respectively meet the judgment probability, and meanwhile, the pixel residual error value information meets the judgment threshold value, the basic block is the regular mark edge part, otherwise, the basic block is not the regular mark edge part; if the intra-frame prediction mode of the basic block belongs to the edge part of the regular mark, the intra-frame 4x4 prediction mode is adopted instead of the intra-frame 16x16 prediction mode, the intra-frame pixel residual error distribution condition CBP value brightness and chroma DC direct current components and AC alternating current components are mostly not 0, the chroma AC alternating current components are basically not 0, the pixel residual error value is larger, and the chroma UV residual error value is more obvious.
Performing binarization processing on all basic blocks, performing noise processing on the binarized basic blocks by adopting morphological opening operation, and reconstructing information of edge parts of the binarized basic blocks due to the opening operation by using closing operation to obtain a rule mark A;
step four, carrying out partial decoding on the inter-frame prediction coding code stream of the video code stream, wherein the partial decoding process comprises entropy decoding and inverse scanning of the inter-frame prediction code stream without inverse quantization, integer IDCT conversion, reconstruction and loop filtering, and obtaining an inter-frame prediction mode, a motion vector residual value and a CBP value of inter-frame pixel residual distribution condition;
and step five, judging and correcting the rule mark A according to the inter-frame prediction mode, the motion vector residual value and the inter-frame pixel residual distribution condition CBP value, and determining whether the rule mark A is a rule mark.
The judgment and correction of the rule mark A refers to the comparison of the shape and the ground color of the rule mark A with the definition of the existing rule mark, the judgment of the type of the mark A carries out partial decoding on the inter-frame prediction coding code stream of the video code stream, the decoding process comprises the entropy decoding and the inverse scanning of the inter-frame prediction code stream, and the inter-frame prediction mode, the motion vector residual value and the inter-frame pixel residual distribution condition CBP value of the basic block are obtained. The method comprises the steps of carrying out judgment and correction on a rule mark A according to an inter-frame prediction mode, a motion vector residual value and an inter-frame pixel residual distribution condition CBP value, wherein the rule mark A comprises basic block prediction modes mostly adopting inter-frame 16x16 prediction modes, the motion vector change is small, most of inter-frame pixel residual conditions CBP values are zero, a statistical rule mark A comprises the number of the basic blocks adopting the inter-frame 16x16 prediction modes, the sum value of absolute values of motion vector residuals and the number of the inter-frame pixel residual conditions CBP values are zero, judgment threshold values are respectively set, meanwhile, the judgment threshold values are met, the rule mark A is a rule mark, and if not, the rule mark A is a pseudo rule mark, and the correction is carried out.
And judging the basic shape of the regular mark A according to the geometric characteristics of the regular mark, wherein the regular mark is mostly in a standard geometric shape such as a circle, a triangle, a rectangle and the like, and the regular mark can be judged by adopting the geometric attributes such as circularity, rectangularity and elongation to obtain the basic shape of the regular mark A. The regular mark has a bright base color, the colors such as blue and red which are obviously different from the surrounding colors are mostly adopted, the chroma residual values of the edge coding blocks of the regular mark are obviously different correspondingly to the code stream, and the base color of the regular mark A can be judged by analyzing the chroma residual value condition. The regular marks of different types are different in shape and background color, the type of the regular marks A can be judged according to the shape and the background color of the regular marks A, if the regular mark indicating marks are circular and blue in background color, and the marks can be judged to be traffic indicating marks when the judging marks are circular and the background color is blue, so that a foundation is laid for further identifying the marks.
The beneficial effects of the invention are as follows: the method comprises the steps of partially decoding an intra-frame prediction code stream of a video code stream to obtain an intra-frame prediction mode, an intra-frame pixel residual error distribution condition CBP value and a pixel residual error value, detecting a rule mark by analyzing the intra-frame prediction mode, the intra-frame pixel residual error distribution condition CBP value and pixel residual error value information, and providing effective information for further identifying the rule mark; the method comprises the steps of obtaining an inter-frame prediction mode, an inter-frame pixel residual error distribution CBP value and a motion vector residual error value by partially decoding an inter-frame prediction code stream of a video code stream, and correcting rule mark detection by analyzing the intra-frame prediction mode, the inter-frame pixel residual error distribution CBP value and the motion vector residual error value information, so that the detection accuracy is improved; the method comprises the steps that the characteristic that the ground colors of different types of rules mark colors are obviously different is utilized, the characteristic that the chroma residual error of an intra-frame prediction coding block in a video code stream has obviously different characteristics is reflected, and the different types of rule marks are identified by utilizing the pixel residual error information of the coding block; the method utilizes the correlation among the pixels of the coding blocks and the residual error information of the coding blocks to carry out detection, thereby avoiding the problem of low detection precision caused by illumination change and improving the detection precision; the method avoids time-consuming operations such as integer IDCT transformation, inverse quantization, reconstruction, loop filtering and the like in the video decoding process, and is beneficial to real-time detection of the rule marks.
Drawings
FIG. 1 is a flow chart of the present invention.
Detailed Description
The technical scheme of the invention is explained in detail by combining the following embodiments: the invention fully utilizes the code stream information such as intra-frame prediction code stream intra-frame prediction mode, intra-frame pixel residual distribution condition CBP value and pixel residual value in the video code stream, inter-frame prediction code stream inter-frame prediction mode, inter-frame pixel residual distribution condition CBP value and motion vector residual value in the video code stream, and the like, and carries out rule mark detection and classification identification on the basis of blocks rather than pixels. The following description will take the rule flag detection and classification recognition as an example.
Assume that the traffic video stream adopts the video compression standard as h.264, and the basic block size is 4x4. The specific detection and classification identification steps are as follows:
the first step is as follows: and partially decoding the intra-frame prediction code stream of the video code stream to obtain the macro block type, the intra-frame prediction mode, the quantization parameter QP, the intra-frame pixel residual error distribution CBP value and the pixel residual error value after partial decoding, wherein time-consuming operations such as integer IDCT transformation, inverse quantization, reconstruction, loop filtering and the like are not required to be completed in the decoding process.
The second step is that: judging the probability P1 of the basic block as a regular mark edge according to the macro block type and the intra-frame prediction mode, and setting P1=0 when the intra-frame prediction mode is an intra-frame 16x16 prediction mode; when the intra prediction mode is the intra 4x4 prediction mode, then P1=1 is set.
The third step: and judging the probability P2 that the basic block is the edge of the regular mark according to the residual error distribution condition CBP value of the pixels in the frame. The intra-frame pixel residual distribution CBP value contains whether a luminance DC coefficient, a luminance AC coefficient, a chrominance DC coefficient and a chrominance AC coefficient are 0 or not, and a pixel residual distribution vector a = [ a1, a2, a3, a4 ] is constructed]The vector contains 4 elements a1, a2, a3 and a4. The luminance DC coefficient is 0, a1=0 is set, otherwise a1=1 is set; the luminance AC coefficient is 0, a2=0 is set, otherwise a2=1 is set; the chroma DC coefficient is 0, setting a3=0, otherwise setting a3=1; the chroma AC coefficient is 0, a4=0, otherwise a4=1 is set. Setting weight vector w of pixel residual distribution 1 ＝[0.2,0.2,0.3,0.3]Calculating the probability P2= a.w 1 T 。
The fourth step: and solving the probability P3= P1xP2, judging that the basic block is not the edge part of the regular mark when P3 is less than 0.5, and turning to the seventh step, otherwise, further judging according to the intra-frame pixel residual value, and turning to the fifth step.
The fifth step: taking the absolute value Y of the DC coefficient of brightness as the pixel residual value DC Absolute value of the chromaticity DC coefficient U DC And V DC SUM of absolute values of AC coefficients of chroma UAC And SUM VAC Setting threshold values TH according to quantization parameters YDC 、TH UDC 、TH VDC 、TH UAC And TH VAC . Construction of a pixel residual value threshold vector b = [ b1, b2, b3, b4, b5 ]]The vector contains 5 elements b1, b2, b3, b4 and b5, where b1= Y DC /TH YDC ,b2＝U DC /TH UDC ,b3＝V DC /TH VDC ,b4＝SUM UAC /TH UAC ,b5＝SUM VAC /TH VAC Setting a pixel residual value threshold weight vector w 2 ＝[0.1,0.2,0.2,0.25,0.25]Calculating a residual judgment threshold TH re ＝b.w 2 T 。
And a sixth step: when TH is re >And when 0.5, judging that the basic block is the edge part of the regular mark, otherwise, judging that the basic block is not the edge part of the regular mark.
The seventh step: and (3) carrying out binarization processing on all basic blocks, wherein the rule mark is assigned with value 255, otherwise, the rule mark is assigned with value 0, carrying out noise processing by adopting morphological opening operation, and reconstructing information of the edge part lost due to the opening operation by using closing operation to obtain a rule mark A.
Eighth step: the basic shapes of the regular marks are circular, triangular and rectangular, the shape matching judgment is carried out on the regular marks A according to the geometric characteristics of the circular, triangular and rectangular shapes, and meanwhile the edges of the regular marks A are corrected according to the shapes of the regular marks.
The ninth step: and judging the background color of the rule mark A according to the residual error threshold condition of the edge basic block of the rule mark A. For example, the base color of the rule flag is blue and is reflected in the UV residual value, the U residual value of the edge portion of the rule flag is large, the base color is red and is reflected in the UV residual value, and the V residual value of the edge portion of the rule flag is large. Suppose that the rule flag A edge contains n 4x4 basic blocks, and each basic block chroma takes chroma DC coefficient absolute value U DCi And V DCi The chroma AC coefficient is SUM which is obtained by summing absolute values of chroma AC coefficients UACi And SUM VACi Threshold value TH for blue decision BDC And TH BAC Red decision threshold TH RDC And TH RAC . When U is turned DCi >TH BDC And U is ACi >TH BAC Judging that the base color of the basic block i of the rule mark is blue when V DCi >TH RDC And V is ACi >TH RAC And judging that the base color of the basic block i is red according to the rule. Counting n 4x4 basic blocks as blue number m when m>(n-3), the rule mark has a blue background color; counting n 4x4 basic blocks as red number m when m>(n-3), the rule indicates that the ground color is red.
The tenth step: and judging which type of the rule mark A is a rule mark warning mark, a prohibition mark, an indication mark and a direction mark according to the shape and the ground color of the rule mark A.
The eleventh step: and performing partial decoding on the inter-frame prediction code stream of the video code stream to obtain the macro block type, the inter-frame prediction mode, the motion vector residual value and the inter-frame pixel residual distribution condition CBP value after partial decoding, wherein time-consuming operations such as pixel residual decoding, integer IDCT transformation, inverse quantization, reconstruction, loop filtering and the like in the decoding process do not need to be completed.
The twelfth step: and (3) taking the center of the block where the regular flag A is located as a search starting point, searching 5 whole-pixel motion vectors from top to bottom, left to right, respectively, searching a region which is most matched with the regular flag A, and taking the minimum value of the sum of absolute values of residual values of the motion vectors as a best matching region B.
The thirteenth step: and counting the probability distribution P of the inter-frame prediction mode of the matching region B. When the inter prediction mode of each 4x4 basic block is SKIP mode or 16x16 mode, the inter prediction mode value M i =1, otherwise M i And =0. Assuming that the matching region B includes n 4 × 4 basic blocks, SUM of inter prediction modes of the statistical matching region B is SUM MODE ＝∑M i Find P = SUM MOD And/n. When P is present<At 0.7, the rule flag is modified, and the rule flag a is a pseudo rule flag but not a rule flag.
The fourteenth step is that: SUM for summing absolute values of residual values of motion vectors in matching region B MVD ，
Wherein, MVD ix For 4x4 basic block transverse motion vector residual, MVD iy The 4x4 basic block longitudinal motion vector residual is obtained, and the ABS is an absolute value operation. When SUM MVD >TH MVD And correcting the rule mark, wherein the rule mark A is a pseudo rule mark and is not a rule mark.
The fifteenth step: and counting the summation value of 0 of the brightness DC coefficient, the brightness AC coefficient, the chroma DC coefficient and the chroma AC coefficient of the matching area B according to the CBP value of the pixel residual distribution condition.
Wherein, Y DC Represents the case of 4x4 basic block luminance DC coefficient, and when it is not 0, Y DC =1, otherwise Y DC ＝0；Y AC Represents the case of 4x4 basic block luminance AC coefficient, and when it is not 0, Y AC =1, otherwise Y AC ＝0；UV DC Representing the case of 4x4 basic block chroma DC coefficient, and when not 0, UV DC =1, otherwise UV DC ＝0；UV AC Represents the case of 4x4 basic block chroma AC coefficient, and is not 0, UV AC =1, otherwise UV AC And =0. When SUM CBP >TH CBP And correcting the rule mark, wherein the rule mark A is a pseudo rule mark and is not a rule mark.
While the invention has been described in further detail in connection with specific embodiments thereof, it will be understood that the invention is not limited thereto, and that various other modifications and substitutions may be made by those skilled in the art without departing from the scope of the invention, which is to be determined by the claims appended hereto.
Claims (1)
1. The method for detecting and classifying and identifying the rule marks based on the video code stream is characterized in that: obtaining an intra-frame prediction mode, an intra-frame pixel residual error distribution CBP value and a pixel residual error value of an intra-frame prediction coding code stream of the video code stream and an inter-frame prediction mode, a motion vector residual error value and an inter-frame pixel residual error distribution CBP value of an inter-frame prediction coding code stream of the video code stream through the obtained video analysis, detecting and classifying the rule marks, and specifically carrying out the following steps
Dividing each frame of image of the obtained video into a plurality of blocks according to 4x4 pixels, wherein each block is defined as a basic block, and carrying out partial decoding on an intra-frame prediction coding code stream of a video code stream, wherein the partial decoding process comprises intra-frame prediction code stream entropy decoding and inverse scanning, and does not carry out inverse quantization, integer IDCT transformation, reconstruction and loop filtering to obtain a macro block type, a quantization parameter, an intra-frame prediction mode, an intra-frame pixel residual error distribution condition CBP value and a pixel residual error value of the basic block;
step two, judging whether the basic block is a regular mark edge part or not according to the intra-frame prediction mode of the basic block, the intra-frame pixel residual error distribution condition CBP value and the pixel residual error value information, if the intra-frame prediction mode of the basic block and the intra-frame pixel residual error distribution condition CBP value respectively meet the judgment probability, and simultaneously the pixel residual error value information meets the judgment threshold value, the basic block is the regular mark edge part, otherwise the basic block is not the regular mark edge part, and the detailed contents are as follows
Judging the probability P1 that the basic block is a regular mark edge according to the macro block type and the intra-frame prediction mode, and setting P1=0 when the intra-frame prediction mode is an intra-frame 16x16 prediction mode; when the intra prediction mode is an intra 4x4 prediction mode, then P1=1 is set;
judging the probability P2 that the basic block is the edge of the regular mark according to the intra-frame pixel residual distribution CBP value, wherein the intra-frame pixel residual distribution CBP value comprises a brightness DC coefficient, a brightness AC coefficient, a chroma DC coefficient and a chroma AC coefficient, and constructing an intra-frame pixel residual distribution vector a = [ a1, a2, a3, a4 ]]The vector contains 4 elements a1, a2, a3 and a4, the luminance DC coefficient is 0, setting a1=0, otherwise setting a1=1; the luminance AC coefficient is 0, a2=0 is set, otherwise a2=1 is set; the chroma DC coefficient is 0, setting a3=0, otherwise setting a3=1; setting a4=0 when the chroma AC coefficient is 0, otherwise setting a4=1, and setting a weight vector w of the pixel residual distribution situation in the frame 1 ＝[0.2,0.2,0.3,0.3]Calculating the probability P2= a.w 1 T ；
Solving the probability P3= P1xP2, when P3 is less than 0.5, judging that the basic block is not the edge part of the regular mark, otherwise, further judging according to the pixel residual value;
taking the absolute value Y of the DC coefficient of brightness as the pixel residual value DC Absolute value of the chromaticity DC coefficient U DC And V DC SUM of absolute values of AC coefficients of chroma UAC And SUM VAC Setting threshold values TH according to quantization parameters YDC 、TH UDC 、TH VDC 、TH UAC And TH VAC Constructing a pixel residual value threshold vector b = [ b1, b2, b3, b4, b5 ]]Vector b contains 5 elements b1, b2, b3, b4 and b5, where b1= Y DC /TH YDC ,b2＝U DC /TH UDC ,b3＝V DC /TH VDC ,b4＝SUM UAC /TH UAC ,b5＝SUM VAC /TH VAC Setting a pixel residual value threshold weight vector w 2 ＝[0.1,0.2,0.2,0.25,0.25]Calculating a pixel residual error judgment threshold TH re ＝b.w 2 T ；
When TH is re >When 0.5, judging the basic block as the edge part of the regular mark, otherwise, judging the basic block not as the edge part of the regular mark;
performing binarization processing on all basic blocks, performing noise processing on the binarized basic blocks by adopting morphological opening operation, and reconstructing information of edge parts of the binarized basic blocks due to the opening operation by using closing operation to obtain a rule mark A;
step four, carrying out partial decoding on the inter-frame prediction coding code stream of the video code stream, wherein the partial decoding process comprises entropy decoding and inverse scanning of the inter-frame prediction code stream without inverse quantization, integer IDCT conversion, reconstruction and loop filtering, and obtaining an inter-frame prediction mode, a motion vector residual value and a CBP value of inter-frame pixel residual distribution condition;
step five, judging and correcting the rule mark A according to an inter-frame prediction mode, a motion vector residual value and an inter-frame pixel residual distribution condition CBP value, determining whether the rule mark A is a rule mark, respectively setting judgment threshold values, and if the judgment threshold values are met, the rule mark A is a rule mark, otherwise, correcting is carried out, the rule mark A is a pseudo rule mark, judging and correcting the rule mark A refers to comparing the shape and the ground color of the rule mark A with the definition of the existing rule mark, judging the type of the rule mark A, searching 5 integer pixel motion vector residual values from top to bottom and from left to right respectively by taking the center of a block where the rule mark A is located as a search starting point as follows, searching a most matching area with the rule mark A, and taking the minimum value of the sum of absolute values of the motion vector residual values as a best matching area B;
counting the probability distribution P of the inter-frame prediction mode in the matching region B, and when the inter-frame prediction mode of each 4x4 basic block is the SKIP mode or the 16x16 mode, determining the value M of the inter-frame prediction mode i =1, otherwise M i =0, matching region B includes n 4x4 basic blocks, statistical matching region B inter prediction mode SUM MODE ＝∑M i Find P = SUM MOD N when P<When 0.7, correcting the rule mark, wherein the rule mark A is a pseudo rule mark and is not a rule mark; SUM of absolute values of residual values of motion vectors in matching region B MVD ，
Wherein the content of the first and second substances,
for the 4x4 basic block lateral motion vector residual,
is the 4x4 basic block longitudinal motion vector residual, ABS is the absolute value operation, when SUM MVD >TH MVD Correcting the rule mark, wherein the rule mark A is a pseudo rule mark and is not a rule mark;
according to the inter-frame pixel residual error distribution CBP value, counting summation values of 0 of a brightness DC coefficient, a brightness AC coefficient, a chroma DC coefficient and a chroma AC coefficient in a matching area B;
wherein, Y DC Represents the case of 4x4 basic block luminance DC coefficient, and when it is not 0, Y DC =1, otherwise Y DC ＝0；Y AC Represents the case of 4x4 basic block brightness AC coefficient, when it is not 0, Y AC =1, otherwise Y AC ＝0；UV DC Representing the case of 4x4 basic block chroma DC coefficient, and when not 0, UV DC =1, otherwise UV DC ＝0；UV AC Represents the case of 4x4 basic block chroma AC coefficient, and is not 0, UV AC =1, otherwise UV AC =0, when SUM CBP >TH CBP And correcting the rule mark, wherein the rule mark A is a pseudo rule mark and is not a rule mark.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011522015.6A CN112714316B (en) | 2020-12-21 | 2020-12-21 | Regular mark detection and classification identification method based on video code stream |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011522015.6A CN112714316B (en) | 2020-12-21 | 2020-12-21 | Regular mark detection and classification identification method based on video code stream |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112714316A CN112714316A (en) | 2021-04-27 |
| CN112714316B true CN112714316B (en) | 2023-01-31 |
Family
ID=75544909
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011522015.6A Active CN112714316B (en) | 2020-12-21 | 2020-12-21 | Regular mark detection and classification identification method based on video code stream |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112714316B (en) |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2006112642A1 (en) * | 2005-04-19 | 2006-10-26 | Samsung Electronics Co., Ltd. | Method and apparatus for adaptively selecting context model for entropy coding |
Family Cites Families (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2012172779A1 (en) * | 2011-06-13 | 2012-12-20 | Panasonic Corporation | Method and apparatus for encoding and decoding video using intra prediction mode dependent adaptive quantization matrix |
| EP2840007B1 (en) * | 2013-08-22 | 2018-04-04 | Honda Research Institute Europe GmbH | Consistent behaviour generation of a predictive advanced driver assistant system |
| AU2014202921B2 (en) * | 2014-05-29 | 2017-02-02 | Canon Kabushiki Kaisha | Method, apparatus and system for de-blocking a block of video samples |
| JP6392572B2 (en) * | 2014-07-22 | 2018-09-19 | ルネサスエレクトロニクス株式会社 | Image receiving apparatus, image transmission system, and image receiving method |
| JP7040466B2 (en) * | 2017-02-13 | 2022-03-23 | ソニーグループ株式会社 | Image processing device and image processing method |
| CN107404653B (en) * | 2017-05-23 | 2019-10-18 | 南京邮电大学 | A kind of Parking rapid detection method of HEVC code stream |
| CN108256511B (en) * | 2018-03-15 | 2022-03-29 | 太原理工大学 | Human motion detection method based on video coding code stream |
| CN108596109B (en) * | 2018-04-26 | 2021-04-13 | 浪潮集团有限公司 | Target detection method and device based on neural network and motion vector |
| CN109379595B (en) * | 2018-11-15 | 2021-10-15 | 太原理工大学 | Dynamic target detection method based on video code stream codebook model |
| JP2022024208A (en) * | 2018-12-07 | 2022-02-09 | シャープ株式会社 | Moving image decoding device and moving image encoding device |
| JP7224892B2 (en) * | 2018-12-18 | 2023-02-20 | ルネサスエレクトロニクス株式会社 | MOVING IMAGE ENCODER AND OPERATION METHOD THEREOF, VEHICLE INSTALLING MOVING IMAGE ENCODER |
| CN111741299B (en) * | 2020-07-09 | 2022-03-25 | 腾讯科技（深圳）有限公司 | Method, device and equipment for selecting intra-frame prediction mode and storage medium |
-
2020
- 2020-12-21 CN CN202011522015.6A patent/CN112714316B/en active Active
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2006112642A1 (en) * | 2005-04-19 | 2006-10-26 | Samsung Electronics Co., Ltd. | Method and apparatus for adaptively selecting context model for entropy coding |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112714316A (en) | 2021-04-27 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Wang et al. | Fuzzy-based algorithm for color recognition of license plates | |
| CN101931815B (en) | Quantization adjustment based on texture level | |
| CN101416511B (en) | Quantization adjustments based on grain | |
| CN111355956B (en) | Deep learning-based rate distortion optimization rapid decision system and method in HEVC intra-frame coding | |
| CN105141948A (en) | Improved HEVC sample point self-adaption compensation method | |
| CN102663358A (en) | Video image identification method of operating state of secondary protection equipment of substation | |
| CN1498391A (en) | Method of and system for detecting cartoon in video data stream | |
| JPH07121656A (en) | Image area separating device | |
| CN1845605A (en) | Image processing and recognizing method for real-time monitoring power high voltage circuit breaker switch state | |
| US20160171314A1 (en) | Unstructured road boundary detection | |
| CN109479131A (en) | Video signal processing method and device | |
| CN111491167B (en) | Image encoding method, transcoding method, device, equipment and storage medium | |
| CN112307953A (en) | Clustering-based adaptive inverse perspective transformation lane line identification method and system | |
| KR102261669B1 (en) | Artificial Neural Network Based Object Region Detection Method, Device and Computer Program Thereof | |
| Chen et al. | A novel fast intra mode decision for versatile video coding | |
| JP4658532B2 (en) | Method for detecting face and device for detecting face in image | |
| CN112437310A (en) | VVC intra-frame coding rapid CU partition decision method based on random forest | |
| US20080253670A1 (en) | Image Signal Re-Encoding Apparatus And Image Signal Re-Encoding Method | |
| CN112714316B (en) | Regular mark detection and classification identification method based on video code stream | |
| Chen et al. | Edge preservation ratio for image sharpness assessment | |
| Liu et al. | Automatic video activity detection using compressed domain motion trajectories for H. 264 videos | |
| Solana-Cipres et al. | Real-time moving object segmentation in H. 264 compressed domain based on approximate reasoning | |
| CN106339995A (en) | Space-time multiple feature based vehicle shadow eliminating method | |
| WO2022116119A1 (en) | Inter-frame predication method, encoder, decoder and storage medium | |
| CN112800968B (en) | HOG blocking-based feature histogram fusion method for identifying identity of pigs in drinking area |
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 | ||
| CP01 | Change in the name or title of a patent holder | ||
| CP01 | Change in the name or title of a patent holder |
Address after: 030001 9-10 / F, East Building, Hongtai international building, 75 Longcheng street, Xiaodian District, Taiyuan City, Shanxi Province Patentee after: Zhilin Information Technology Co.,Ltd. Address before: 030001 9-10 / F, East Building, Hongtai international building, 75 Longcheng street, Xiaodian District, Taiyuan City, Shanxi Province Patentee before: Taiyuan Zhilin Information Technology Co.,Ltd. |