Algorithmic and architectural design for real-time and power-efficient Retinex image/video processing

Abstract

This paper presents novel algorithmic and architectural solutions for real-time and power-efficient enhancement of images and video sequences. A programmable class of Retinex-like filters, based on the separation of the illumination and reflectance components, is proposed. The dynamic range of the input image is controlled by applying a suitable non-linear function to the illumination, while the details are enhanced by processing the reflectance. An innovative spatially recursive rational filter is used to estimate the illumination. Moreover, to improve the visual quality results of two-branch Retinex operators when applied to videos, a novel three-branch technique is proposed which exploits both spatial and temporal filtering. Real-time implementation is obtained by designing an Application Specific Instruction-set Processor (ASIP). Optimizations are addressed at algorithmic and architectural levels. The former involves arithmetic accuracy definition and linearization of non-linear operators; the latter includes customized instruction set, dedicated memory structure, adapted pipeline, bypasses, custom address generator, and special looping structures. The ASIP is synthesized in standard-cells CMOS technology and its performances are compared to known Digital signal processor (DSP) implementations of real-time Retinex filters. As a result of the comparison, the proposed algorithmic/architectural design outperforms state-of-art Retinex-like operators achieving the best trade-off between power consumption, flexibility, and visual quality

Filtering Algorithms for Image Processing in Embedded Real-time Systems

“…The algorithm is substantially similar to FAST, SURF and Harris algorithms in detecting and handling image noise, but its advantages and highlights are more prominent. In embedded real-time image processing system, with the decrease of SNR (Saponara et al, 2007) (Signal to Noise Ratio, in both video playback and image displaying) and the increase of fuzzy degree on after another, all the methods based on grayscale will gradually fail to work, algorithms including FAST, SURF and Harris, although bound to a certain false detection rate, have better robust performance against Gauss Noise and Salt and Pepper Noise. Below is a comparative analysis of Kalman Motion Filtering Algorithm and FAST, SURF and Harris Algorithms in terms of related ability to eliminate possible Gauss Noise and Salt and Pepper noise in images.…”

Real-Time Biologically-Inspired Image Exposure Correction

“…The first has been implemented on a Digital Signal Processor (DSP) [4,5], allowing the real-time, single-scale rendering of grayscale images, with sizes up to 256×256 pixels. A variation of the second has been implemented on an Application Specific Instruction-set Processor (ASIP) [6], allowing the processing of SXGA (1280×768 pixels) or WXGA (1366×768 pixels) still images in 1 sec, or the a real-time rendering of video frames with size 256×256 pixels and frame rates up to 29 frames per second (fps). Both implementations do not meet the VGA standard (color images of 640×480 pixels size and 25fps) in video rendering.…”

Fully-Integrated Converter for Low-Cost and Low-Size Power Supply in Internet-of-Things Applications

“…These "things" can be part of complex systems for healthcare [18][19][20][21][22][23][24][25], smart agriculture [26], science experiments [27], vehicles [28][29][30][31][32], satellites [33][34][35], domotics [36,37], robots/drones or industrial machines [38][39][40], telecom or surveillance apparatus [41][42][43], sport [44,45] or energy/smart-grid systems [46][47][48][49][50], and/or consumer electronics [51][52][53][54][55][56][57]. These units are usually wireless nodes, working at sub-GHz frequencies or in the range 2.4 GHz to 6 GHz [58][59][60][61], or at mm-waves [62][63][64].…”