Deep neural network (NN) has been widely accepted for enabling various AI applications, however, the limitation of computational and memory resources is a major problem on mobile devices. Quantized NN with a reduced bit precision is an effective solution, which relaxes the resource requirements, but the accuracy degradation due to its numerical approximation is another problem. We propose a novel quantized NN model employing the “dithering” technique to improve the accuracy with the minimal additional hardware requirement at the view point of the hardware-algorithm co-designing. Dithering distributes the quantization error occurring at each pixel (neuron) spatially so that the total information loss of the plane would be minimized. The experiment we conducted using the software-based accuracy evaluation and FPGA-based hardware resource estimation proved the effectiveness and efficiency of the concept of an NN model with dithering.

A digital background calibration technique using signal-dependent dithering is proposed, to correct the nonlinear errors which results from capacitor mismatches and finite opamp gain in pipelined analog-to-digital converter (ADC). Large magnitude dithers are used to measure and correct both errors simultaneously in background. In the proposed calibration system, the 2.5-bit capacitor-flip-over multiplying digital-to-analog converter (MDAC) stage is modified for the injection of large magnitude dithering by adding six additional comparators, and thus only three correction parameters in every stage subjected to correction were measured and extracted by a simple calibration algorithm with multibit first stage. Behavioral simulation results show that, using the proposed calibration technique, the signal-to-noise-and-distortion ratio improves from 63.3 to 79.3dB and the spurious-free dynamic range is increased from 63.9 to 96.4dB after calibrating the first two stages, in a 14-bit 100-MS/s pipelined ADC with σ=0.2% capacitor mismatches and 60dB nonideal opamp gain. The time of calibrating the first two stages is around 1.34 seconds for the modeled ADC.

This letter presents an improved visible watermarking scheme for halftone images. It incorporates watermark embedding into ordered dither halftoning by threshold modulation. The input images include a continuous-tone host image (e.g. an 8-bit gray level image) and a binary watermark image, and the output is a halftone image with a visible watermark. Our method is content adaptive because it takes local intensity information of the host image into account. Experimental results demonstrate effectiveness of the proposed technique. It can be used in practical applications for halftone images, such as commercial advertisement, content annotation, copyright announcement, etc.

The artifacts of low-bit rate quantization in images cannot be removed satisfactorily by known methods. We propose decomposition of images as HSI and LSI (higher- and lower- significance images), followed by subsampling and reconstruction methods for LSI. Experiments show significant improvement in image quality, as compared to other methods.

This paper proposes a new blocking artifact reduction algorithm using an adaptive filter based on classifying the block boundary area. Generally, block-based coding, such as JPEG and MPEG, introduces blocking and ringing artifacts to an image, where the blocking artifact consists of grid noise, staircase noise, and corner outliers. In the proposed method, staircase noise and corner outliers are reduced by a 1D low-pass filter. Next, the block boundaries are divided into two classes based on the gradient of the pixel intensity in the boundary region. For each class, an adaptive filter is applied so that the grid noise is reduced in the block boundary regions. Thereafter, for those blocks with an edge component, the ringing artifact is removed by applying an adaptive filter around the edge. Finally, high frequency components are added to those block boundaries where the natural characteristics have been lost due to the adaptive filter. The computer simulation results confirmed a better performance by the proposed method in both the subjective and objective image qualities.

We proposed and demonstrated a novel wavelength stabilization technique for dense wavelength division multiplexing (DWDM) systems using dithering-induced AM cancellation which improves both wavelength stability and data transmission performance. Our wavelength stabilization technique consists of an optical frequency discriminating function and a function for canceling AM components induced by frequency dithering of the light source. The frequency discrimination in this technique is based on an FM-AM conversion effect, obtained by interaction from frequency dithering of the light with the bandpass characteristic of an arrayed-waveguide grating (AWG) multiplexer. The AM cancellation function was added to suppress optical frequency discriminating errors occurring due to AM components induced by frequency dithering in this wavelength stabilization architecture. In this scheme, an electro-absorption (EA) modulator is employed not only for modulating high-speed data traffic but also for suppressing AM components induced by frequency dithering on the light signal. Since the EA modulator is usually used for modulating high-speed data traffic, dedicated optical devices are not required for suppressing the AM components. The wavelength stability of a light source can therefore be enhanced with simple architecture. In the demonstration, a reduction of fluctuations within 50 MHz versus changes in the modulation index, and long-term stability within 320 MHz after more than 60 hours was achieved in 10 Gbit/s NRZ transmission. We also confirmed that the proposed AM cancellation technique effectively reduces the transmission penalties due to frequency dithering in 10 Gbit/s NRZ data transmission performance.

We proposed and demonstrated a novel wavelength stabilization technique for dense wavelength division multiplexing (DWDM) systems using dithering-induced AM cancellation which improves both wavelength stability and data transmission performance. Our wavelength stabilization technique consists of an optical frequency discriminating function and a function for canceling AM components induced by frequency dithering of the light source. The frequency discrimination in this technique is based on an FM-AM conversion effect, obtained by interaction from frequency dithering of the light with the bandpass characteristic of an arrayed-waveguide grating (AWG) multiplexer. The AM cancellation function was added to suppress optical frequency discriminating errors occurring due to AM components induced by frequency dithering in this wavelength stabilization architecture. In this scheme, an electro-absorption (EA) modulator is employed not only for modulating high-speed data traffic but also for suppressing AM components induced by frequency dithering on the light signal. Since the EA modulator is usually used for modulating high-speed data traffic, dedicated optical devices are not required for suppressing the AM components. The wavelength stability of a light source can therefore be enhanced with simple architecture. In the demonstration, a reduction of fluctuations within 50 MHz versus changes in the modulation index, and long-term stability within 320 MHz after more than 60 hours was achieved in 10 Gbit/s NRZ transmission. We also confirmed that the proposed AM cancellation technique effectively reduces the transmission penalties due to frequency dithering in 10 Gbit/s NRZ data transmission performance.