RESUMO
This paper presents a low-voltage low-power chopper-stabilized differential difference amplifier (DDA) realized using 40 nm CMOS technology. Operating with a supply voltage of 0.5 V, a three-stage DDA has been employed to achieve an open-loop gain of 89 dB, while consuming just 0.74 µW of power. The proposed DDA incorporates feed-forward frequency compensation and a Type II compensator to achieve pole-zero cancellation and damping factor control. The DDA has a unity-gain bandwidth (UGB) of 170 kHz, a phase margin (PM) of 63.98°, and a common-mode rejection ratio (CMRR) of up to 100 dB. This circuit can effectively drive a 50 pF capacitor in parallel with a 300 kΩ resistor. The use of the chopper stabilization technique effectively mitigates the offset and 1/f noise. The chopping frequency of the chopper modulator is 5 kHz. The input noise is 245 nV/sqrt (Hz) at 1 kHz, and the input-referred offset under Monte Carlo cases is only 0.26 mV. Such a low-voltage chopper-stabilized DDA will be very useful for analog signal processing applications. Compared to the reported chopper DDA counterparts, the proposed DDA is regarded as that with one of the lowest supply voltages. The proposed DDA has demonstrated its effectiveness in tradeoff design when dealing with multiple parameters pertaining to power consumption, noise, and bandwidth.
RESUMO
Due to distribution shift, deep learning based methods for image dehazing suffer from performance degradation when applied to real-world hazy images. In this paper, this study considers a dehazing framework based on conditional diffusion models for improved generalization to real haze. First, our work finds that optimizing the training objective of diffusion models, i.e., Gaussian noise vectors, is non-trivial. The spectral bias of deep networks hinders the higher frequency modes in Gaussian vectors from being learned and hence impairs the reconstruction of image details. To tackle this issue, this study designs a network unit, named Frequency Compensation block (FCB), with a bank of filters that jointly emphasize the mid-to-high frequencies of an input signal. Our work demonstrates that diffusion models with FCB achieve significant gains in both perceptual and distortion metrics. Second, to further boost the generalization performance, this study proposed a novel data synthesis pipeline, HazeAug, to augment haze in terms of degree and diversity. Within the framework, a solid baseline for blind dehazing is set up where models are trained on synthetic hazy-clean pairs, and directly generalize to real data. Extensive evaluations on real dehazing datasets demonstrate the superior performance of the proposed dehazing diffusion model in distortion metrics. Compared to recent methods pre-trained on large-scale, high-quality image datasets, our model achieves a significant PSNR improvement of over 1 dB on challenging databases such as Dense-Haze and Nh-Haze.