Gidel’s lossless Image Compression IP targets FPGA-based Imaging & Vision applications. This lossless image compression IP performs real-time compression for Color Filter Array (CFA – e.g., Bayer), Monochrome, and RGB images and videos. It supports multi-camera/sensor systems at pixel clock rates exceeding 1 Gpixel/s, using minimal FPGA resources and power. In real-world video tests, it achieves a 1:2.3 compression ratio at 8 bits/pixel. Higher bit configurations are available upon request.
The IP integrates seamlessly with Gidel’s comprehensive eco-system, enabling tailored solutions with image processing, vision algorithms, and concurrent recording. Recording can also be enhanced by Gidel’s CamSim playback system.
Two compression modes are available:
Frame compression – compress individual images.
Video compression – compress video using I and P frames.
Compression can be optimized via pre-training on sample image data or on-the-fly training from real-time data.
Gidel’s Lossless compression is highly modular.
Want to learn more?
| Name | Description | Type | Size | |
|---|---|---|---|---|
|
Lossless Compression
|
Datasheet | 176.9 KB | ||
|
Lossless Compression
|
System Block Diagram: Compression in the FPGA data flow | PNG | 207.9 KB | |
|
Lossless Compression
|
Performance Table | PNG | 56.6 KB | |
|
Lossless Compression: Article
|
Gidel Launches Lossless Compression IP that Reduces Storage Needs by Over 50%, Utilizing Only 1%... | Link | — | Read article |
The primary advantage is the ability to perform real-time, bit-perfect compression for high-bandwidth imaging. It ensures that the original data can be 100% reconstructed without a single bit of loss. This is critical for medical, scientific, or defense applications where data integrity is the top priority, all while achieving an average 1:2.3 compression ratio (at 8 bits/pixel).
The IP is designed to support camera rates exceeding 1 Gpixel/s and is fully supported by the Gidel InfiniVision acquisition system. This allows for synchronized, real-time compression from multiple cameras or sensors simultaneously, making it ideal for large-scale, high-throughput vision architectures.
The Lossless IP is modular and can be deployed across Gidel’s full hardware range, including FantoVision Edge Computers, PCIe Frame Grabbers, PCIe FPGA Accelerators, and ultra-compact FPGA modules. Because it is highly optimized (typically requiring less than 1,000 ALMs), it leaves ample FPGA resources for other vision algorithms.
Designers can choose between two distinct modes to fit their workflow:
Frame Compression: Compresses individual images (ideal for triggered or sparse acquisition).
Video Compression: Compresses continuous streams using I and P frames to maximize bandwidth efficiency during long-duration recording.
Yes. Unlike some compression methods that favor one over the other, Gidel’s Lossless IP is specifically designed to perform reliably for both moving and static cameras. It ensures consistent data reduction regardless of the motion within the frame.
Compression can be optimized through two methods: pre-training on sample image data provided by the user, or on-the-fly training using real-time data from the live stream. This flexibility ensures the highest possible compression ratio for the specific image characteristics of your application.
