Friday, November 20, 2026
Science
No Result
View All Result
  • Login
  • HOME
  • SCIENCE NEWS
  • CONTACT US
  • HOME
  • SCIENCE NEWS
  • CONTACT US
No Result
View All Result
Scienmag
No Result
View All Result
Home Science News

Fast Prediction Algorithm Cuts Spatial Scalable Video Coding Time by Seventy-Nine Percent

November 20, 2026
in Science News
Reading Time: 6 mins read
0
Fast Prediction Algorithm Cuts Spatial Scalable Video Coding Time by Seventy-Nine Percent

Fast Prediction Algorithm Cuts Spatial Scalable Video Coding Time by Seventy-Nine Percent

65
SHARES
587
VIEWS
Share on FacebookShare on Twitter
ADVERTISEMENT

A research team has demonstrated an optimization method that reduces the intra coding time of Spatial Scalable High Efficiency Video Coding by an average of 78.64 percent while incurring a negligible coding efficiency change of a 0.27 percent decrease in BDBR. The framework, introduced in a study published in the EURASIP Journal on Image and Video Processing by Weian Li, Weihua Ou, Hongbing Wang, Yang Wu, and Yunhao Zhong, accelerates the encoding pipeline by coupling structural coding unit partitions directly with mode selection probabilities. Scalable video architectures are increasingly demanded by digital television broadcasting, video conferencing, wireless streaming, and mobile communications, where terminal devices feature diverse display resolutions. While Spatial Scalable High Efficiency Video Coding provides an effective multi-layer approach to adapt video streams to varying resolutions, its multi-layer structure and inter-layer prediction mechanisms introduce severe computational complexity that has historically hindered practical, widespread deployment.

Spatial Scalable High Efficiency Video Coding organizes compressed video streams into a hierarchical structure comprising a Base Layer and one or more Enhancement Layers. The Base Layer adheres to standard High Efficiency Video Coding prediction routines, whereas the Enhancement Layers incorporate inter-layer prediction to maximize compression efficiency across spatial layers. Because standard High Efficiency Video Coding is already computationally demanding, the recursive splitting and evaluation required across multiple layers compound the computational burden. Standard configurations recursively evaluate 64-by-64 pixel Coding Tree Units down to 8-by-8 pixel Coding Units. In Enhancement Layers, encoders must repeatedly evaluate both the Inter-layer Reference mode and Intra mode across thirty-five directional modes to identify the configuration with the lowest Rate Distortion cost. To bypass unnecessary computations, the authors designed a three-stage decision framework: Textural Features-Based Coding Units Prediction, Probability-Based Intra Mode Early Skip, and Probability and Distribution-Based Direction Mode Selection.

The first stage of the framework accelerates Coding Unit partitioning by analyzing pixel texture properties rather than computing brute-force evaluations for every potential sub-division. Coding Units containing simple textures generally settle on smaller coding depths such as 64-by-64 or 32-by-32 pixels, while complex textures demand deeper sub-divisions such as 16-by-16 or 8-by-8 pixels. The researchers selected a Decision Tree classifier over Support Vector Machines or Naive Bayes models due to its transparent flow-chart-like structure, low operational overhead, and high classification accuracy. To train the Decision Tree, the investigators extracted eigenvalues capturing overall pixel variance, horizontal and vertical half-block variances, and quadrant-based pixel variances. In addition, four directional Sobel operator components oriented at zero, forty-five, ninety, and one hundred thirty-five degrees measured directional complexity across three-by-three pixel neighborhoods. Combined with the Quantization Parameter, these eigenvalues allow the encoder to estimate the probability that a Coding Unit is optimal. If the calculated probability is at least 0.9, recursive evaluation terminates immediately; if the probability is 0.1 or lower, checking for that Coding Unit is skipped entirely.

The second stage of the pipeline addresses the high computational cost of mode selection within Enhancement Layer Coding Units. Under standard operation, Enhancement Layer blocks evaluate both Inter-layer Reference mode and Intra mode. Empirical measurements conducted by the research team across benchmark video sequences revealed that Inter-layer Reference mode accounts for an average of only 19.18 percent of total coding time while representing 73.75 percent of the optimal selections. Conversely, Intra mode demands 80.82 percent of the coding time but is chosen as optimal in only 26.25 percent of instances. Recognizing this substantial asymmetry, the researchers established an early-skipping mechanism that tests the computationally inexpensive Inter-layer Reference mode first. A Decision Tree evaluates textural variance, directional complexity, neighboring Coding Unit modes, and Rate Distortion cost to determine the probability of the Inter-layer Reference mode. Because Intra mode is nested inside its collocated Coding Unit, its final selection probability is computed by multiplying the derived Intra probability by the probability of its collocated Coding Unit. When this joint probability falls at or below 0.1, the encoder skips Intra mode evaluation entirely.

For Coding Units that do not qualify for an early Intra mode skip, the third stage adaptively streamlines the evaluation of the thirty-five candidate Direction Modes. Exhaustively evaluating all directional possibilities incurs severe computational overhead in standard encoders. By analyzing the statistical selection frequency of Direction Modes across 8-by-8 Coding Units, the authors found that the ten most frequently chosen modes account for 77.79 percent of all optimal selections, whereas the remaining twenty-five modes represent just 22.21 percent. Designating these top ten modes as the Basic Modes Set, the algorithm scales the candidate pool according to the probability of the collocated Coding Unit. When the collocated Coding Unit probability is 0.25 or lower, the encoder restricts its search strictly to the ten modes of the Basic Modes Set, which experimental data confirmed maintains an approximately constant BDBR. As the Coding Unit probability rises, the search set expands incrementally: four additional high-probability modes are added when probability is below or equal to 0.35, another four when below or equal to 0.45, and six more when below or equal to 0.55. For Coding Units exceeding a 0.55 probability, the system employs gradient amplitude sums derived from Sobel operators to predict potential modes without exhaustively testing unlikely candidates.

The experimental evaluation was conducted using the Spatial Scalable High Efficiency Video Coding reference software, SHM version 11.0, executing on a server configured with an Intel 2.0 GHz central processing unit and 30 gigabytes of memory. Training and testing adhered to the Common SHM Test Conditions across eight representative video sequences: Blue_sky, Ducks, Park_Joy, Pedestrian, Tractor, Town, Station2, and Average. These sequences were deliberately chosen to span diverse motion profiles, varying texture complexities, and different degrees of inter-layer correlation. The authors tested two spatial scalability ratios, 1.5x and 2x, representing the dimensional scaling factor of frame height and width between the Base Layer and Enhancement Layer. The Base Layer was encoded using a fixed Quantization Parameter set of 22, 26, 30, and 34, while the Enhancement Layer was tested under two separate Quantization Parameter sets: 22, 26, 30, 34 and 24, 28, 32, 36. Across all four resulting experimental configurations, performance was quantified via the BDBR metric for coding efficiency and the percentage of time saving in the Enhancement Layer for coding speed.

Evaluating the three individual algorithmic components under the most demanding testing condition, Case 4, demonstrated how each strategy contributes to the overall speedup. Textural Features-Based Coding Units Prediction achieved an average coding time saving of 43.10 percent with an average BDBR change of negative 0.04 percent. Probability-Based Intra Mode Early Skip produced the largest individual speed improvement, reaching a 58.79 percent time saving alongside a 0.19 percent increase in BDBR, reflecting the major time savings achieved by eliminating complex Intra mode calculations. The Probability and Distribution-Based Direction Mode Selection module delivered a 27.41 percent time saving with a negative 0.16 percent BDBR change. Because the early termination thresholds applied during Coding Unit partitioning are strict, that stage yielded virtually no compression degradation, whereas skipping Intra mode provided substantial acceleration at the cost of a minor increase in Rate Distortion cost.

When combining all three strategies into an integrated framework, the authors benchmarked their method against two state-of-the-art fast intra coding algorithms, PBFIP and HSEIP, across all four test cases. In Case 1, corresponding to a 1.5x scalability ratio with identical Base Layer and Enhancement Layer Quantization Parameters, the proposed algorithm achieved a 77.60 percent time saving with a 0.56 percent BDBR, outperforming PBFIP’s 57.59 percent time saving and HSEIP’s 72.17 percent time saving. In Case 2, which maintained a 1.5x ratio with higher Enhancement Layer Quantization Parameters, the algorithm achieved a 77.77 percent time saving and a 0.43 percent BDBR, compared to 60.25 percent for PBFIP and 73.86 percent for HSEIP. In Case 3, utilizing a 2x scalability ratio, the method recorded a 79.26 percent time saving with a 0.10 percent BDBR, exceeding PBFIP by 17.43 percentage points and HSEIP by 5.01 percentage points. In Case 4, featuring a 2x ratio and mismatched Quantization Parameters, the algorithm achieved a 79.92 percent time saving with a 0.0 percent BDBR, surpassing PBFIP’s 64.34 percent and HSEIP’s 75.95 percent.

Across all four experimental configurations, the integrated algorithm delivered an overall average coding speed improvement of 78.64 percent while maintaining an overall average BDBR of 0.27 percent. By comparison, PBFIP achieved an average speed improvement of 61.00 percent with a negative 0.03 percent BDBR, and HSEIP achieved 74.06 percent with a 0.21 percent BDBR. The proposed algorithm demonstrated a speed advantage of 17.64 percentage points over PBFIP and 4.58 percentage points over HSEIP. The authors observed that BDBR shifts were more pronounced under the 1.5x scalability ratio than under the 2x ratio. In the 2x configurations, the wider spatial resolution gap between the Base Layer and Enhancement Layer diminished the precision of inter-layer interpolation, rendering the Inter-layer Reference mode less dominant and altering the relative performance balance.

Performance variations across individual sequences revealed how spatial texture, inter-layer dependency, and object motion interact within the prediction model. Sequences with complex textures, such as ParkScene, experienced higher gains from the texture-based Coding Unit prediction stage, recording a 43.26 percent time saving compared to 41.12 percent for more homogeneous material like BQTerrace, as rich textural variance provides clear decision boundaries for the Decision Tree. Inter-layer dependency also proved critical: sequences with strong cross-layer correlations like Kimono achieved time savings up to 84.27 percent in Case 4, whereas sequences with weaker dependencies such as BasketballDrive showed lower gains of 76.10 percent due to reduced collocated prediction reliability. Furthermore, rapid motion introduced higher performance variability, illustrated by BasketballDrive exhibiting a 1.5 percent BDBR shift in Case 1 compared to 0.2 percent in Case 4, because fast movement disrupted both inter-layer alignment and directional mode selections. The researchers concluded that jointly linking mode probabilities to collocated Coding Units significantly reduces redundant processing, and they noted plans to examine deep learning classifiers in future research.

Subject of Research: Fast intra prediction algorithm for spatial scalable high efficiency video coding

Article Title: Jointly probability-based fast intra prediction algorithm for spatial SHVC

Article References: Li, W., Ou, W., Wang, H., Wu, Y., & Zhong, Y. (2025). Jointly probability-based fast intra prediction algorithm for spatial SHVC. EURASIP Journal on Image and Video Processing, 2025(1), Article 20. https://doi.org/10.1186/s13640-025-00683-2

Image Credits: AI Generated

DOI: 10.1186/s13640-025-00683-2

Keywords: Spatial Scalable High Efficiency Video Coding, SSHVC, Intra prediction, Coding Units, Decision Tree, BDBR, Rate Distortion cost, Direction Modes, Inter-layer prediction, Base Layer, Enhancement Layer, Sobel Operator, Quantization Parameter, Coding complexity, Video compression, Common SHM Test Conditions

Cite Scienmag News

Scienmag. (November 20, 2026). Fast Prediction Algorithm Cuts Spatial Scalable Video Coding Time by Seventy-Nine Percent. https://scienmag.com/fast-prediction-algorithm-cuts-spatial-scalable-video-coding-time-by-seventy-nine-percent/

Scienmag. "Fast Prediction Algorithm Cuts Spatial Scalable Video Coding Time by Seventy-Nine Percent." Scienmag, 20 November 2026, https://scienmag.com/fast-prediction-algorithm-cuts-spatial-scalable-video-coding-time-by-seventy-nine-percent/. Accessed 20 November 2026.

Scienmag. "Fast Prediction Algorithm Cuts Spatial Scalable Video Coding Time by Seventy-Nine Percent." Scienmag. November 20, 2026. https://scienmag.com/fast-prediction-algorithm-cuts-spatial-scalable-video-coding-time-by-seventy-nine-percent/

Tags: Base LayerBDBRCoding complexityCoding UnitsCommon SHM Test Conditionsdecision treeDirection ModesEnhancement LayerInter-layer predictionIntra predictionQuantization ParameterRate Distortion costSobel OperatorSpatial Scalable High Efficiency Video CodingSSHVCvideo compression
Share26Tweet16
Previous Post

European Port Responses to Investment Screening Depend Heavily on Local Governance

Next Post

Smart Radiative Textiles Expand Options for Energy-Efficient Personal Thermal Management

Related Posts

Multicenter ICU Study Identifies Three Sociodemographic Profiles Among Relatives Providing Surrogate Consent
Science News

Multicenter ICU Study Identifies Three Sociodemographic Profiles Among Relatives Providing Surrogate Consent

November 20, 2026
High Life’s Essential 8 Scores Associated With Lower Mortality in Early CKM
Science News

High Life’s Essential 8 Scores Associated With Lower Mortality in Early CKM

November 20, 2026
Targeting Folate Receptor Overcomes Immunotherapy Resistance in Colorectal Cancer in Mice
Cancer

Targeting Folate Receptor Overcomes Immunotherapy Resistance in Colorectal Cancer in Mice

November 20, 2026
Jiangxi Cities Narrowed Gaps in Healthcare Workforce and Beds Over Ten Years
Medicine

Jiangxi Cities Narrowed Gaps in Healthcare Workforce and Beds Over Ten Years

November 20, 2026
Dimethoate Metabolites Pose Greater Predicted Toxicological Risks Than Parent Insecticide
Medicine

Dimethoate Metabolites Pose Greater Predicted Toxicological Risks Than Parent Insecticide

November 20, 2026
Dual-Frequency Electroacupuncture Modulates Neurotransmitters and Relieves Insomnia in Mice
Medicine

Dual-Frequency Electroacupuncture Modulates Neurotransmitters and Relieves Insomnia in Mice

November 20, 2026
Next Post
Smart Radiative Textiles Expand Options for Energy-Efficient Personal Thermal Management

Smart Radiative Textiles Expand Options for Energy-Efficient Personal Thermal Management

  • Mothers who receive childcare support from maternal grandparents show more optimized

    Mothers who receive childcare support from maternal grandparents show more parental warmth, finds NTU Singapore study

    27656 shares
    Share 11059 Tweet 6912
  • University of Seville Breaks 120-Year-Old Mystery, Revises a Key Einstein Concept

    1061 shares
    Share 424 Tweet 265
  • Bee body mass, pathogens and local climate influence heat tolerance

    682 shares
    Share 273 Tweet 171
  • Researchers record first-ever images and data of a shark experiencing a boat strike

    546 shares
    Share 218 Tweet 137
  • Groundbreaking Clinical Trial Reveals Lubiprostone Enhances Kidney Function

    531 shares
    Share 212 Tweet 133
Science

Embark on a thrilling journey of discovery with Scienmag.com—your ultimate source for cutting-edge breakthroughs. Immerse yourself in a world where curiosity knows no limits and tomorrow’s possibilities become today’s reality!

RECENT NEWS

  • Multicenter ICU Study Identifies Three Sociodemographic Profiles Among Relatives Providing Surrogate Consent
  • High Life’s Essential 8 Scores Associated With Lower Mortality in Early CKM
  • Targeting Folate Receptor Overcomes Immunotherapy Resistance in Colorectal Cancer in Mice
  • Jiangxi Cities Narrowed Gaps in Healthcare Workforce and Beds Over Ten Years

Categories

  • Agriculture
  • Anthropology
  • Archaeology
  • Athmospheric
  • Biology
  • Biotechnology
  • Blog
  • Bussines
  • Cancer
  • Chemistry
  • Climate
  • Earth Science
  • Editorial Policy
  • Marine
  • Mathematics
  • Medicine
  • Pediatry
  • Policy
  • Psychology & Psychiatry
  • Science Education
  • Science News
  • Social Science
  • Space
  • Technology and Engineering

Subscribe to Blog via Email

Enter your email address to subscribe to this blog and receive notifications of new posts by email.

Join 5,151 other subscribers

© 2026 Scienmag - Science Magazine

Welcome Back!

Login to your account below

Forgotten Password?

Retrieve your password

Please enter your username or email address to reset your password.

Log In
No Result
View All Result
  • HOME
  • SCIENCE NEWS
  • CONTACT US

© 2026 Scienmag - Science Magazine

Discover more from Science

Subscribe now to keep reading and get access to the full archive.

Continue reading