Pixel Cluster Compression for Low-Power Image Encoding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional compression algorithms impose a high computational load due to frequency domain transformations, leading to increased power consumption and inefficient data compression and decompression processes.
Innovation Solution
Parent-child cluster (PCC) compression technique groups pixels in RGB space into clusters, utilizing a clustering method that reduces dimensionality and employs a key encoding scheme tailored to the human visual system, providing high fidelity in low entropy regions and high dynamic range in high entropy regions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If conventional compression algorithms use frequency domain transformations, then compression ratio is improved, but computational load and power consumption increase
Solution Approach 1:
The patent replaces frequency domain transformations (mathematical/mechanical system) with spatial domain clustering operations. Instead of using DCT or wavelet transforms that require complex calculations, the invention uses simple distance-based cluster assignments in RGB space, significantly reducing computational load while maintaining compression effectiveness
Solution Approach 2:
The patent changes the domain of operation from frequency domain to spatial domain, and from continuous color space to discrete cluster centroids. This parameter transformation allows for simpler quantization operations that reduce power consumption while achieving comparable compression ratios
2Loss of information
If conventional compression algorithms use frequency domain transformations, then compression ratio is improved, but computational complexity increases
Solution Approach 1:
The patent substitutes complex frequency domain transformation algorithms with simple spatial clustering operations. The cluster assignment process uses straightforward distance calculations in RGB space rather than complex transform matrices, dramatically reducing computational complexity
Solution Approach 2:
The patent segments the image into clusters based on spatial proximity in color space. By dividing the continuous color space into discrete clusters and assigning pixels to nearest centroids, the algorithm achieves compression without requiring complex global transformations
3Measurement precision
If cluster centroids are updated with each pixel assignment, then clustering accuracy is improved, but processing time increases
Solution Approach 1:
The patent performs preliminary cluster centroid calculations using a subset of pixels before final assignment. By pre-establishing centroids based on initial sampling, the algorithm reduces the computational burden during final pixel assignment while maintaining clustering accuracy
Solution Approach 2:
The patent updates cluster centroids for only some pixels or uses incremental updates rather than recalculating all centroids with all pixels. This partial action approach maintains sufficient clustering accuracy while significantly reducing processing time
Data Source
AI summary
A system includes a memory and a display. The system also includes a processor coupled to the memory and to the display. The processor is configured to obtain an image and determine a parent cluster of pixels of an image having a centroid. The processor is also configured to split the parent cluster into at least a first child cluster and a second child cluster and assign a pixel of the image to the first child cluster. Additionally, the pixel is configured to update a centroid of the first child cluster responsive to the pixel and replace the pixel in the image with the centroid of the first child cluster to produce a compressed image. Also, the processor is configured to store the compressed image in the memory, where the display is configured to display the compressed image.