Derivative works — US Patent 8175148

Defensive Disclosure: Quantization Parameter Indication in Video Coding Systems

This document outlines derivative variations of US Patent 8175148, focusing on methods and devices for indicating quantization parameters in a video coding system. The aim is to preemptively disclose enhancements that extend the scope and application of the core invention, rendering future incremental improvements obvious or non-novel. The original patent introduces the concept of a sequence-level quantization parameter (SQP) and transmitting difference values (ΔQP) for individual pictures/slices, thereby reducing bit-rate.

Core Claims Targeted for Derivation:

Independent Claim 1 (Method of Encoding)
Independent Claim 13 (Method of Decoding)
Independent Claim 22 (Encoder Device)
Independent Claim 33 (Decoder Device)

Derivative Variations

1. Material & Component Substitution

Derivative 1.1: Alternative Transform and Quantization Functions

Enabling Description: Instead of a Discrete Cosine Transform (DCT) and simple scalar quantization, the system employs a Wavelet Transform (WT) for energy compaction, specifically the 9/7 biorthogonal wavelet, followed by vector quantization (VQ). The SQP would then represent a default VQ codebook index or a scaling factor applied to a base VQ codebook. ΔQP would indicate a deviation from this default, such as a shift to an alternative codebook from a predefined set, or an additive/multiplicative adjustment to the chosen codebook's vectors. This maintains the principle of a sequence-level default and differential signaling for localized adaptation.
Targeted Claims: Independent Claims 1, 13, 22, 33.

Mermaid Diagram:

graph TD
    A[Input Video] --> B{Wavelet Transform};
    B --> C{Vector Quantizer (VQ)};
    C --> D{SQP/ΔQP Logic};
    D -- SQP/ΔQP --> E[Encoded Bitstream];
    E --> F{SQP/ΔQP Reconstruct};
    F -- Reconstructed VQ Params --> G{Inverse Vector Quantizer};
    G --> H{Inverse Wavelet Transform};
    H --> I[Output Decoded Video];

Derivative 1.2: Neural Network-Based Adaptive Quantization Parameter Generation

Enabling Description: The quantization parameter (QP), including both SQP and ΔQP, is dynamically determined and predicted by a compact neural network (NN) integrated within the encoder. The NN takes as input various video characteristics (e.g., motion complexity, texture, scene change detection, bit-rate buffer fullness) and outputs a predicted optimal QP. The SQP could be a learned global bias for the NN's output for the sequence, and ΔQP would represent a smaller, differential adjustment learned locally by the NN for each slice/frame, trained to minimize rate-distortion costs. The NN parameters themselves or a seed for their generation could be the "SQP" with ΔQP as a minor refinement or explicit override for specific conditions.
Targeted Claims: Independent Claims 1, 13, 22, 33.

Mermaid Diagram:

graph TD
    A[Input Video Features] --> B{Neural Network QP Predictor};
    B -- SQP (learned bias) --> C{QP Generation Logic};
    B -- ΔQP (local adjustment) --> C;
    C --> D[Quantizer];
    D --> E[Encoded Bitstream];
    E --> F[Decoder NN QP Predictor];
    F --> G[Inverse Quantizer];
    G --> H[Output Decoded Video];

Derivative 1.3: GPU-Accelerated Quantization and Inverse Quantization

Enabling Description: The computationally intensive quantization and inverse quantization operations, including the SQP/ΔQP calculation and application, are offloaded to a dedicated Graphics Processing Unit (GPU) or a specialized hardware accelerator (e.g., FPGA or ASIC). The GPU's parallel processing capabilities allow for rapid adjustment and application of QP across numerous macroblocks or blocks simultaneously. The control logic for deriving the actual QP from SQP and ΔQP (e.g., adding ΔQP to SQP) is executed on the GPU, maximizing throughput for high-resolution, high-frame-rate video streams.
Targeted Claims: Independent Claims 22, 33, 44, 45.

Mermaid Diagram:

graph TD
    A[CPU (Control Manager)] --> B{GPU/FPGA Accelerator};
    B -- Frame/Slice Data --> C[Quantizer/Inverse Quantizer Kernel];
    A -- SQP, ΔQP Input --> B;
    C -- Quantized/Inverse Quantized Data --> D[Output];
    B -- Reconstructed QP --> A;

2. Operational Parameter Expansion

Derivative 2.1: Ultra-Low Latency, High-Frequency QP Updates

Enabling Description: For real-time interactive applications (e.g., cloud gaming, remote surgery), the system supports QP updates at sub-macroblock granularity or multiple times per macroblock row, with the SQP defined for very short bursts (e.g., a group of pictures, GOP, of 1-5 frames) rather than an entire sequence. ΔQP signaling can occur at every macroblock or even 4x4 block boundary to enable highly granular quality adaptation in response to rapid changes in content or network conditions, minimizing latency. This requires highly optimized parsing and application of ΔQP values to avoid overhead.
Targeted Claims: Independent Claims 1, 13, 22, 33.

Mermaid Diagram:

sequenceDiagram
    Encoder->>+Decoder: Transmit SQP (per GOP)
    loop Per Frame
        Encoder->>Decoder: Transmit ΔQP (per Slice/MB row)
        Decoder->>Decoder: Reconstruct QP
        Decoder->>Decoder: Inverse Quantize
    end

Derivative 2.2: Extreme-Scale Video Coding (e.g., 16K Resolution / Microscopic Imagery)

Enabling Description: The system is adapted for video sequences at extreme resolutions (e.g., 15360x8640 pixels, 16K) or for microscopic imaging with very fine detail. The SQP is set as a baseline for the entire massive frame, while ΔQP is critically used to adapt quantization within highly localized, perceptually important regions (Regions of Interest - ROIs) that might span only a few macroblocks. For instance, in scientific imaging, an ROI might be a specific cell structure, requiring finer quantization (lower QP, negative ΔQP) than background areas. The system must efficiently manage and signal a large number of ΔQP values without overwhelming the bit-rate.
Targeted Claims: Independent Claims 1, 13, 22, 33.

Mermaid Diagram:

graph LR
    A[16K Video Frame] --> B{Frame Partitioning};
    B --> C[Background Region];
    B --> D[ROI 1];
    B --> E[ROI 2];
    C -- SQP --> F{Quantizer};
    D -- SQP + ΔQP_ROI1 --> F;
    E -- SQP + ΔQP_ROI2 --> F;
    F --> G[Encoded Bitstream];

Derivative 2.3: Dynamic Range Quantization with Logarithmic QP Scaling

Enabling Description: For High Dynamic Range (HDR) video content, the system employs a logarithmic scaling of the quantization parameter (QP) rather than a linear one. The SQP is defined in a logarithmic domain (e.g., log₂(QP_base)) and transmitted. ΔQP values are also represented logarithmically, such that the effective QP applied is QP = 2^(SQP_log + ΔQP_log). This allows for more precise control over quantization levels across a wider range of pixel intensities, especially critical in very dark or very bright regions of HDR content where small changes in QP have a significant perceptual impact.
Targeted Claims: Independent Claims 1, 13, 22, 33.

Mermaid Diagram:

flowchart TD
    A[Input HDR Video] --> B{Logarithmic QP Calculation};
    B -- SQP_log --> C[Encoder: Quantization];
    B -- ΔQP_log --> C;
    C --> D[Encoded Bitstream];
    D --> E{Decoder: Inverse Quantization};
    E -- SQP_log --> F[Reconstruct QP: 2^(SQP_log + ΔQP_log)];
    E -- ΔQP_log --> F;
    F --> G[Output Decoded HDR Video];

3. Cross-Domain Application

Derivative 3.1: Medical Imaging Data Compression (MRI/CT Scans)

Enabling Description: The SQP/ΔQP mechanism is applied to the compression of volumetric medical imaging data, such as MRI or CT scans. Here, a "sequence" could be a 3D volume or a time-series of 3D volumes. The SQP defines a default quantization level for the entire scan, while ΔQP is used to signal finer quantization in diagnostically critical regions (e.g., tumors, lesions) or specific anatomical slices, ensuring high fidelity where needed most, while reducing overall data size for storage and transmission. The "blocks" would correspond to 3D sub-volumes (voxels).
Targeted Claims: Independent Claims 1, 13, 22, 33.

Mermaid Diagram:

graph TD
    A[3D Medical Volume] --> B{3D Transform (e.g., 3D DCT/Wavelet)};
    B --> C{Define SQP for Volume};
    C -- SQP --> D{Quantize Critical Region};
    D -- ΔQP_Critical --> D;
    C -- SQP --> E{Quantize Non-Critical Region};
    D --> F[Encoded Stream];
    E --> F;

Derivative 3.2: Geospatial Satellite Imagery Compression

Enabling Description: The SQP/ΔQP concept is used for compressing multi-spectral or hyperspectral satellite imagery, where a "sequence" could be a geographical region captured over time or a single high-resolution image with many spectral bands. The SQP sets a baseline quantization for common land features. ΔQP is then employed to preserve detail in specific areas of interest, such as urban development zones, agricultural fields undergoing rapid change, or disaster areas, where precise spectral and spatial information is crucial. Each spectral band or sub-region could have its own ΔQP.
Targeted Claims: Independent Claims 1, 13, 22, 33.

Mermaid Diagram:

graph LR
    A[Satellite Image (Multi-band)] --> B{Per-Band Transform};
    B --> C{Define SQP (Global)};
    C -- SQP --> D{Quantize General Terrain};
    D -- ΔQP_Urban/Agri --> D;
    C --> E{Quantize ROI (e.g., Disaster Area)};
    E -- ΔQP_Disaster --> E;
    D --> F[Compressed Imagery];
    E --> F;

Derivative 3.3: Financial Time-Series Data Compression

Enabling Description: Applied to the compression of high-frequency financial time-series data (e.g., stock ticks, order book changes) for archival or low-latency transmission. A "sequence" would be a continuous stream of market data over a trading day. The SQP establishes a default quantization precision for common price and volume fluctuations. ΔQP is then used to signal higher precision (lower quantization) during periods of high volatility, significant market events, or for specific assets that require utmost accuracy, thereby preserving critical information without excessive storage/bandwidth demands during quiescent periods. Transform coding could involve wavelets or other spectral decompositions of the time series.
Targeted Claims: Independent Claims 1, 13, 22, 33.

Mermaid Diagram:

sequenceDiagram
    participant Market_Feed_Generator
    participant Encoder_System
    participant Decoder_System
    participant Archive_System

    Market_Feed_Generator->>Encoder_System: Raw Time-Series Data
    Encoder_System->>Encoder_System: Apply Transform & Define SQP (Day Basis)
    loop Per Time Interval
        Encoder_System->>Encoder_System: Detect Volatility/Event
        alt High Volatility/Event
            Encoder_System->>Encoder_System: Calculate ΔQP (for higher precision)
        else Low Volatility
            Encoder_System->>Encoder_System: Calculate ΔQP (for lower precision)
        end
        Encoder_System->>Encoder_System: Quantize Data (SQP + ΔQP)
        Encoder_System->>Decoder_System: Transmit Encoded Data (incl. ΔQP)
        Decoder_System->>Decoder_System: Reconstruct QP (SQP + ΔQP)
        Decoder_System->>Decoder_System: Inverse Quantize
        Decoder_System->>Archive_System: Decoded Data
    end

4. Integration with Emerging Tech

Derivative 4.1: AI-Driven Predictive SQP/ΔQP Optimization with IoT Feedback

Enabling Description: An AI model (e.g., Reinforcement Learning agent) dynamically optimizes the SQP and subsequent ΔQP values based on real-time feedback from IoT sensors. For example, in a smart city surveillance system, an SQP might be set based on expected traffic density. If IoT environmental sensors detect adverse weather (e.g., fog, heavy rain) or acoustic sensors detect an anomaly (e.g., crash), the AI could trigger a localized, temporary decrease in QP (negative ΔQP) for affected camera feeds, prioritizing critical details in challenging conditions, while maintaining SQP for stable scenes. The SQP itself can be updated by the AI based on long-term trends or policy changes.
Targeted Claims: Independent Claims 1, 13, 22, 33, 44, 45.

Mermaid Diagram:

graph TD
    A[IoT Sensors (Weather, Sound)] --> B{AI Optimizer};
    C[Video Encoder] --> B;
    B -- Optimal SQP --> C;
    B -- Optimal ΔQP (Real-time) --> C;
    C --> D[Encoded Stream];
    D --> E[Video Decoder];
    E --> F[Output Display];
    B -- Feedback Loop --> G[System Monitoring];
    G --> B;

Derivative 4.2: Blockchain-Verified Quantization Parameters for Tamper-Proof Archiving

Enabling Description: For applications requiring undeniable proof of video integrity and quality settings (e.g., legal evidence, journalistic archives), the SQP for a video sequence and subsequent ΔQP values for frames/slices are cryptographically hashed and recorded on a blockchain. Before encoding, the encoder calculates the SQP and a hash of it is committed to a blockchain transaction. Each ΔQP value transmitted in the bit-stream is also accompanied by a hash, or a hash of a group of ΔQPs, which links back to the SQP hash on the blockchain. The decoder can then verify the authenticity and integrity of the QP settings by comparing locally computed hashes against the blockchain record, ensuring no tampering with the quantization parameters.
Targeted Claims: Independent Claims 1, 13, 22, 33, 44, 45.

Mermaid Diagram:

sequenceDiagram
    participant Encoder
    participant Blockchain
    participant Decoder

    Encoder->>Encoder: Determine SQP
    Encoder->>Blockchain: Commit hash(SQP)
    Blockchain-->>Encoder: Transaction ID
    loop Per Slice/Frame
        Encoder->>Encoder: Determine ΔQP
        Encoder->>Encoder: Include ΔQP, hash(ΔQP) in bitstream
        Encoder->>Decoder: Transmit bitstream
        Decoder->>Decoder: Extract SQP, ΔQP, hashes
        Decoder->>Blockchain: Retrieve hash(SQP) (using Transaction ID)
        Blockchain-->>Decoder: Stored hash(SQP)
        Decoder->>Decoder: Verify hash(SQP) and hash(ΔQP)
        alt Verification Success
            Decoder->>Decoder: Reconstruct QP (SQP + ΔQP)
        else Verification Failure
            Decoder->>Decoder: Flag Tampering
        end
    end

Derivative 4.3: Edge Computing Assisted QP Adaptation with Contextual Awareness

Enabling Description: The video encoding and decoding occur at the edge of the network (e.g., on smart cameras, local gateways). An edge processing unit collects local contextual data (e.g., local network congestion, available computational resources, immediate environment changes). This contextual data is used to inform the AI model (as in Derivative 4.1) or a heuristic engine that, instead of sending a global SQP, calculates a "contextual SQP" relevant to the edge device's immediate operating conditions. ΔQP is then applied for further micro-adjustments within the frame/slice, but the base SQP is re-evaluated frequently based on edge context, leading to highly localized and adaptive quality control optimized for edge deployments.
Targeted Claims: Independent Claims 1, 13, 22, 33, 44, 45.

Mermaid Diagram:

graph LR
    A[Edge Camera/Sensor] --> B{Contextual Data Collector (Network, Compute, Environment)};
    B --> C{SQP/ΔQP Adaptation Engine (Edge)};
    C -- Contextual SQP --> D[Encoder (Edge)];
    C -- ΔQP --> D;
    D --> E[Encoded Stream];
    E --> F[Decoder (Edge/Cloud)];
    F --> G[Output];

5. The "Inverse" or Failure Mode

Derivative 5.1: Graceful Degradation in Low-Power/Limited-Functionality Mode

Enabling Description: When operating under severe power constraints (e.g., battery-powered IoT device) or limited processing capability, the system enters a "low-power mode." In this mode, the SQP is automatically increased (coarser quantization) across the entire sequence to significantly reduce bit-rate and computational load. Furthermore, ΔQP signaling is restricted to a very limited range (e.g., only allowing small positive ΔQP to increase QP further, or only at frame level, not slice level) to simplify parsing and decision-making. The encoder might explicitly signal this mode, or the decoder infers it from the received SQP and ΔQP range limits.
Targeted Claims: Independent Claims 1, 13, 22, 33, 44, 45.

Mermaid Diagram:

stateDiagram-v2
    [*] --> Normal_Operation
    Normal_Operation --> Low_Power_Mode: Power/Resource Constraint
    Low_Power_Mode --> Normal_Operation: Power/Resource Available

    state Normal_Operation {
        Normal_Operation : Full SQP/ΔQP Range
        Normal_Operation : High Fidelity Encoding/Decoding
    }

    state Low_Power_Mode {
        Low_Power_Mode : Increased SQP (Coarser)
        Low_Power_Mode : Restricted ΔQP Range/Frequency
        Low_Power_Mode : Reduced Computational Load
    }

Derivative 5.2: Error Concealment Prioritization with Quantization Bias

Enabling Description: In lossy transmission environments (e.g., wireless networks), the SQP/ΔQP mechanism is enhanced to aid error concealment. The SQP can be biased to be slightly lower (finer quantization) for I-frames or key frames, making them more robust. For P-frames or B-frames, ΔQP values for perceptually critical regions (e.g., faces, moving objects) are encoded with higher redundancy or are explicitly biased towards finer quantization (lower QP), while less important regions can tolerate higher QPs. If a ΔQP value is lost, the decoder defaults to the SQP or a "safe" ΔQP (e.g., zero) for that region, reducing visual artifacts more gracefully than a random error.
Targeted Claims: Independent Claims 1, 13, 22, 33.

Mermaid Diagram:

graph TD
    A[Input Video] --> B{Error Resilient Encoder};
    B -- I-Frame SQP_I, ΔQP_I --> C[Quantizer];
    B -- P/B-Frame SQP_P, ΔQP_P(Critical) --> C;
    B -- P/B-Frame SQP_P, ΔQP_P(Non-Critical) --> C;
    C --> D[Encoded Bitstream (Error-Resilient)];
    D --> E{Lossy Channel};
    E --> F{Error Concealing Decoder};
    F --> G[Output Decoded Video (Reduced Artifacts)];

Derivative 5.3: Region-of-Interest (ROI) Failure Mode with Fixed SQP

Enabling Description: In situations where the system needs to guarantee a minimum quality for a designated Region of Interest (ROI) even during severe bandwidth constraints or processing limitations, the SQP is defined globally for the entire video, but the ΔQP mechanism is primarily used to increase quantization (coarsen) for non-ROI regions. For the ROI, ΔQP is always fixed to a value that ensures a minimum acceptable quality (effectively, a lower QP). If the system cannot meet the target bit-rate with this setting, it first increases ΔQP for non-ROI regions, and only as a last resort increases the global SQP, thus protecting the ROI's quality. This provides a "safe" failure mode where only non-critical areas degrade first.
Targeted Claims: Independent Claims 1, 13, 22, 33.

Mermaid Diagram:

graph LR
    A[Input Frame] --> B{ROI Detector};
    B -- ROI Mask --> C{QP Control Logic};
    C -- Global SQP --> D[Quantizer];
    C -- ΔQP_NonROI (Adjustable) --> D;
    C -- ΔQP_ROI (Fixed Low) --> D;
    D --> E[Encoded Bitstream];
    E --> F[Decoder];
    F --> G[Output Decoded Video (ROI Protected)];

Combination Prior Art Scenarios

These scenarios combine the method and device of US Patent 8175148 with existing open-source video coding standards, making the present invention's principles obvious within these contexts.

H.264/MPEG-4 AVC (Advanced Video Coding) Standard:
- Description: The H.264 standard widely utilizes quantization parameters (qp_init_minus2 at the sequence parameter set (SPS) or picture parameter set (PPS) level, and slice_qp_delta or mb_qp_delta at slice or macroblock level). The inventive concept of US8175148 directly maps to this. An H.264 encoder could define an SQP equivalent to a base qp_init_minus2 in the SPS. Subsequent slice_qp_delta or mb_qp_delta values would then serve as the ΔQP, which are added to the effective picture-level QP (derived from the base qp_init_minus2 and any PPS-level pic_init_qp_minus2). This combination shows that the explicit signaling of a sequence-level QP and differential QPs at lower granularities is a natural and obvious extension for bit-rate reduction in an H.264 compliant encoder/decoder.
- Reference: ITU-T Rec. H.264 (03/2005) / ISO/IEC 14496-10:2005, particularly sections on Sequence Parameter Set (SPS), Picture Parameter Set (PPS), and Slice Header syntax.
VP9 Video Codec (Open-source by Google):
- Description: VP9, an open and royalty-free video coding format, employs a base_q_idx (base quantizer index) at the frame level and allows for delta_q_lf (delta for loop filter quantizer) or per-segment q_index_delta values. The concept of US8175148 can be directly applied. The base_q_idx could serve as the SQP for a group of frames (similar to a sequence), and the q_index_delta values for different segments within a frame would function as ΔQPs. This demonstrates the patent's core idea is directly transferable to a modern, open-source codec design, where a common reference (SQP) is established, and local adjustments (ΔQP) are signaled differentially.
- Reference: VP9 Bitstream Specification, specifically sections related to "Quantizer Index" and "Quantization Parameters," typically found in official VP9 documentation.
AV1 Video Codec (Alliance for Open Media):
- Description: AV1, also an open and royalty-free video coding format, uses a base_q_idx similar to VP9, but with more sophisticated quantization parameter signaling options, including delta_q_y_dc, delta_q_uv_dc, and delta_q_uv_ac for different transform types and color components, potentially signaled per-segment or per-tile. The US8175148's invention, specifying an SQP (e.g., the initial base_q_idx for a segment or tile group) and then signaling subsequent delta_q values as ΔQP, is a straightforward adaptation. An AV1 implementation would define a sequence-level SQP in a metadata header, and then use the existing delta_q syntax within frames/tiles/segments as the ΔQP, reconstructing the actual QP based on the received differential values, thereby enhancing bit-rate efficiency for QP signaling within the AV1 framework.
- Reference: AV1 Bitstream Specification, specifically chapters on "Quantization Parameters" and "Frame Header," available from the Alliance for Open Media (AOMedia).

USPTO Search for US8175148:

The USPTO website provides tools for searching patents, such as Patent Public Search (PPUBS) and Patent Center. To search for a specific patent number like US8175148, one would use the "Patent or Publication number" field in the Basic search interface of Patent Public Search. The provided instructions indicate that for patent numbers with 7 digits, no leading zeros are needed if it already has 7 digits, but for 6 digits or less, leading zeros should be added to make it 7 total digits. Since US8175148 is a 7-digit patent number, it would be entered directly as "8175148" into the search tool. While the search results describe how to use the USPTO search tools, they do not directly return the patent document itself or any additional information beyond what was provided in the prompt. Therefore, to provide specific details about US8175148 from the USPTO database, I would need to perform the search using the actual USPTO tools. However, based on the authoritative full patent text provided at the beginning, I already have the necessary information for US8175148.The USPTO website provides tools for searching patents, such as Patent Public Search (PPUBS) and Patent Center. To search for a specific patent number like US8175148, one would use the "Patent or Publication number" field in the Basic search interface of Patent Public Search. As US8175148 is a 7-digit patent number, it would be entered directly as "8175148" into the search tool. The search results describe how to use these tools but do not directly return the patent document or any additional information beyond what was initially provided in the authoritative full patent text.