Low-energy Pipelined Hardware Design for Approximate Medium Filter

Abstract

The image and video processing algorithms are currently crucial for many applications. Hardware implementation of these algorithms provides higher speed for large computation applications. Besides, noise removing is often a typical pre-processing step to enhance the results of later analysis and processing. Median filter is a typical nonlinear filter that is very commonly used for impulse noise elimination in digital image processing. This paper suggests a low-energy median filter hardware design for battery based hardware applications. Approximate solution with high accuracy is investigated to speed up the filtering operation, reduce the area, and consume less power/energy. Pipelining and parallelism are used to optimize the speed and power of this technique. Non-pipelined, two different pipelined structures, and two parallel architectures versions are designed. The design versions are implemented firstly with a Virtex-5 LX110T FPGA then using the UMC 130nm standard cells ASIC technology. The selection and the even-odd sorting-based median filters are also implemented for an equitable comparison with the standard median filtering techniques. The suggested non-pipelined median filter design enhances the throughput by 35% than the highest investigated state-of-the-art. The pipelining enhances the throughput to more than twice its value. Additionally, the parallel architecture decreases the area and the consumed power by around 40%. The simulation results reveal that one of the suggested designs significantly decreases the area, with the same speed as the fastest design in literature without noticeable degrading the accuracy, and a significant decrease in energy consumption by about 60%.

Autoencoder-based Image Denoiser Suitable for Image of Numbers with High Potential for Hardware Implementation

“…In contrast, the most traditional AI based method is training the neural network using pairs of data containing clean and noisy versions of the same image to detect the noise and replace the corrupted pixels with the correct ones from the clean set. Therefore, many approaches for removing the noise from images using neural networks like convolutional neural networks with median layer, noise2noise image denoising, generative adversarial networks, and autoencoders [16]. These approaches are still being developed to denoise the image and increase the image quality.…”

“…MSE and PSNR are inversely proportional; a lower MSE means higher image quality after denoising. Figure 5 illustrates the CNN median layer, which consists of 64 layers; between each layer, a median layer acts as a median filter [16]. [16] To test this network, a dataset of 90 200x200-pixel photos was used to train the model.…”

“…Figure 5 illustrates the CNN median layer, which consists of 64 layers; between each layer, a median layer acts as a median filter [16]. [16] To test this network, a dataset of 90 200x200-pixel photos was used to train the model. The network's output is a 70x70-pixel scaled image that contains the image's denoised version.…”