Lossless Image Compression Using Variable-Bit Pixel Difference Encoding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image data compression techniques, such as JPEG, are resource-intensive and difficult to implement in Field Programmable Gate Arrays (FPGAs), especially for lossless compression and decompression of image data in video cameras, which require efficient compression methods to handle various protocols and high-speed data transmission.
Innovation Solution
A lossless image data compression method that encodes differences between pixel values using the smallest number of bits required, with a header structure that includes bits to represent the number of bits per pixel difference and additional information, allowing for efficient compression and decompression of image data, even in hardware implementations like FPGAs.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If lossless compression algorithms are implemented in FPGA, then data compression capability is improved, but device complexity and resource consumption increase
Solution Approach 1:
The image data is divided into multiple line segments, with each segment processed independently using its own header structure. This segmentation allows the compression algorithm to be implemented in modular fashion in FPGA, reducing overall device complexity while maintaining compression capability.
Solution Approach 2:
The patent dynamically changes the number of bits per pixel (t) based on the actual difference values in each segment. By using variable-bit encoding where t is an integer between 1 and N, the system adapts to local data characteristics, improving compression efficiency without requiring complex fixed-structure implementations.
2Loss of substance
If traditional JPEG compression is used, then compression ratio is improved, but implementation difficulty and resource requirements increase
Solution Approach 1:
The patent extracts only the essential difference information between consecutive line segments, encoding merely the changes rather than full pixel data. This extraction approach achieves compression similar to JPEG but with much simpler hardware implementation, as it avoids complex transform and quantization operations.
Solution Approach 2:
Instead of compressing by transforming image data into frequency domain (as in JPEG), this patent inverts the approach by directly encoding spatial differences between adjacent lines in the spatial domain. This inversion simplifies the compression mechanism while maintaining effective data size reduction.
3Productivity
If variable bit depth encoding is used, then compression efficiency is improved, but processing complexity increases
Solution Approach 1:
The patent performs preliminary determination of the bit depth parameter t for each segment before actual encoding. By pre-calculating the required bits based on maximum difference values in each segment and storing this in the header, the system avoids complex real-time variable-bit processing during encoding, reducing operational complexity.
Solution Approach 2:
The system dynamically adjusts the bit depth parameter t for each line segment based on actual data characteristics. This dynamic adaptation allows optimal compression efficiency for each segment while the parameter is determined once per segment, balancing efficiency gains with manageable processing complexity.
Data Source
AI summary
An apparatus for compressing a video image file is disclosed. The image file includes line segments having a length of M pixels of N bits and a header of H bits. Each pixel is represented by t bits. The apparatus includes a circuit configured to read a first number of unencoded pixel values in a first line segment and read a second number of unencoded pixel values of a second line segment are read. The first and second line segments preceded by a header of H bits and each pixel value is represented by t bits. The differences between each of the first and second unencoded pixel values is determined. The differences between each of the first and second unencoded pixel values encoded only using a smallest number of bits t for a given number of bits for each pixel value in the segment, wherein t is defined as an integer between 1 and N.