Complexity-adaptive Image Compression via Prediction Error Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional image compression schemes require a large number of complex context models to achieve an acceptable compression ratio, making them difficult to implement and inefficient in processing.
Innovation Solution
The technique separates the prediction error of image data into distinct factors (sign, bit category, and relative absolute value) and applies a separate set of context models to each factor, using binary representations that adapt based on the size of the prediction error to simplify the compression process.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a large number of context models are applied to achieve acceptable compression ratio, then compression performance is improved, but algorithm complexity and implementation difficulty increase
Solution Approach 1:
The prediction error is segmented into multiple independent factors: sign, bit category, and relative absolute value. Each factor is then processed with its own dedicated context models, reducing the complexity of any single processing unit while maintaining overall compression effectiveness through the combined processing of all factors.
Solution Approach 2:
Different context models are applied to different factors based on their specific characteristics. The sign factor uses one set of context models, the bit category factor uses another set, and the relative absolute value factor uses a third set. This localized approach allows optimization for each factor's specific properties without overwhelming complexity.
2Reliability
If a large number of context models are applied to achieve acceptable compression ratio, then compression performance is improved, but processing efficiency deteriorates
Solution Approach 1:
By dividing the compression task into separate factor processing stages, each stage can operate independently with optimized context models. This segmentation enables parallel processing and reduces the computational burden on any single processing unit, thereby improving overall processing efficiency while maintaining compression ratio.
Solution Approach 2:
The system dynamically selects and applies context models based on the specific characteristics of each factor being processed. This dynamic adaptation allows the system to optimize processing efficiency for each factor type rather than using a fixed, overly complex set of models for all cases.
3Reliability
If conventional compression schemes are used, then compression ratio can be achieved, but implementation difficulty increases
Solution Approach 1:
The implementation is segmented into distinct modules, each handling a specific factor (sign processing module, bit category processing module, relative absolute value processing module). This modular segmentation makes the system easier to implement, debug, and maintain while achieving the desired compression ratio through the coordinated operation of these modules.
Solution Approach 2:
Each factor receives dedicated context models tailored to its specific properties, simplifying the implementation requirements for each module. The localized approach allows for more straightforward coding and easier maintenance compared to a monolithic system that must handle all complexity uniformly.
Data Source
AI summary
An improved technique of compressing image data involves separating a prediction error of image data into distinct factors and applying a separate set of context models to each factor. Such factors may take the form of a sign, a bit category, and a relative absolute value of the prediction error. For each factor, the improved technique provides a set of context models and a procedure for selecting a context model from each respective set. The context model for each factor determines a probability distribution of symbols that may represent that factor, which in turn enables compression of the prediction error. Additionally, the symbols that represent certain factors into which the prediction error is separated result from a binary representation whose form—either unary or uniform—depends on the size of the prediction error.