An Optimized HW/SW Implementation of the Vector Median Rational Hybrid Filter for Real-Time Color Image Denoising
Bibliographic record
Abstract
The presence of noise in an image can significantly diminish its visual quality and adversely affect the accuracy of subsequent image processing tasks.Therefore, it is imperative to enhance image quality in real-time by eliminating disturbances introduced during the image acquisition or transmission process.This paper proposes an efficient and optimized implementation of the vector median rational hybrid filter (VMRHF) specifically tailored for real-time color image denoising.This filter is crafted to harness the capabilities of both the vector median filter and the rational operator, enabling effective noise reduction while maintaining the integrity of edges, image details, and chromaticity.However, the hybrid architecture in the VMRHF filter brings about an increase in computational complexity.To address this complexity, the filter is implemented in a Hardware/Software (HW/SW) codesign context, capitalizing on the strengths of both hardware and software components.The software component is created using the C/C++ programming language and operates on the ARM Cortex-A53 processor with a clock frequency of 1.2 GHz, while the high-level synthesis (HLS) flow is employed to develop the hardware portion, implemented as a coprocessor in the Zynq UltraScale+ XCZU9EG FPGA.Nevertheless, in the pursuit of crafting an optimized hardware architecture for VMRHF, specific directives like ARRAY_PARTITION and PIPELINE are progressively incorporated into the VMRHF C code using the Xilinx Vivado HLS tool.The interaction between the hardware and software parts is streamlined through the AXI-stream interface, facilitated by three Direct Memory Access (DMA) units for efficient data parallel transfer, thereby boosting data throughput.The VMRHF HLS design is evaluated on the embedded ZCU102 kit.The experimental outcomes illustrate that our design is capable of restoring a 256256 color image within 19 ms, reflecting a substantial 94% decrease in execution time compared to the software design.This notable improvement is achieved while upholding consistent image quality, as indicated by both objective measures such as peak signal-to-noise ratio (PSNR) and subjective assessments.These results hold true across various levels of "salt and pepper" impulsive noise.Besides, our design exhibits a power consumption of merely 4.46 watts.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.001 |
| Open science | 0.001 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one teacher head, not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".