Image Encoding Apparatus Subband Code Distribution
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image encoding techniques face challenges in maintaining image quality while achieving a target encoded data amount, particularly when dealing with RAW moving images, as they often result in block distortion and deterioration due to inappropriate setting of quantization parameters across subbands and planes.
Innovation Solution
An image encoding apparatus that performs frequency conversion using discrete wavelet transform, followed by quantization and encoding, with a code amount controlling unit that determines subband target code amounts based on a difficulty indicator to distribute the target code amount effectively across subbands, thereby maintaining image quality and achieving the desired encoded data amount.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If DCT-based compression encoding is used to compress moving images, then compression efficiency is improved, but block distortion occurs and image quality deteriorates at high compression rates
Solution Approach 1:
The patent changes the fundamental encoding parameter from DCT (Discrete Cosine Transform) to DWT (Discrete Wavelet Transform). This parameter change allows the system to achieve high compression ratios while avoiding block distortion, as DWT provides better energy compaction and frequency localization properties compared to DCT, thereby resolving the contradiction between compression efficiency and image quality preservation.
Solution Approach 2:
The patent segments the image into multiple subbands through wavelet decomposition, separating different frequency components (LL, LH, HL, HH subbands). This segmentation allows differential processing and quantization of different frequency regions, enabling efficient compression while preserving important visual information and avoiding uniform block artifacts across the entire image.
2Productivity
If quantization parameter Qp is increased to reduce code amount, then compression rate is improved, but image quality deteriorates
Solution Approach 1:
The patent applies local quality by setting different quantization parameters for different subbands based on their visual importance. Important subbands (such as LL and low-frequency components) use smaller Qp values to preserve quality, while less important high-frequency subbands use larger Qp values for greater compression. This localized differentiation resolves the contradiction by allowing aggressive compression in non-critical regions while maintaining quality in visually significant areas.
3Ease of operation
If target code amount is set equally for all subbands, then encoding simplicity is maintained, but Qp ratio deviates from predetermined relationship and image quality deteriorates
Solution Approach 1:
The patent performs preliminary action by pre-calculating and setting appropriate target code amounts for each subband based on their visual importance and statistical characteristics, rather than setting them equally. This preliminary differentiation ensures that the predetermined Qp ratio relationship is maintained throughout encoding, preventing quality deterioration while keeping the process systematic and manageable.
Solution Approach 2:
The patent implements feedback mechanisms to monitor the actual code amounts generated for each subband and adjust quantization parameters accordingly to maintain the predetermined Qp ratio relationships. This feedback control ensures that the target code amount distribution is achieved in practice, resolving the deviation issue that would otherwise occur with equal target code amount settings.
Data Source
AI summary
An image encoding apparatus of the present invention comprises a frequency converting unit which executes frequency conversion on the data of a frame in moving image data, and generates coefficient data of a plurality of subbands, an encoding unit which quantizes the coefficient data acquired by the frequency converting unit in accordance with a quantization parameter, and encodes the quantized coefficient data, and a code amount controlling unit which controls the encoding unit such that a code amount generated by the encoding unit achieves a target code amount are provided. Here, the code amount controlling unit determines a subband target code amount of each of the plurality of subbands by distributing the target code amount to the plurality of subbands based on a ratio determined according to an indicator indicating a difficulty calculated for each of the plurality of subbands.