Upsampling

About: Upsampling is a research topic. Over the lifetime, 2426 publications have been published within this topic receiving 57613 citations.

Upsampling Artifacts in Neural Audio Synthesis

Abstract: A number of recent advances in neural audio synthesis rely on up-sampling layers, which can introduce undesired artifacts. In computer vision, upsampling artifacts have been studied and are known as checkerboard artifacts (due to their characteristic visual pattern). However, their effect has been overlooked so far in audio processing. Here, we address this gap by studying this problem from the audio signal processing perspective. We first show that the main sources of upsampling artifacts are: (i) the tonal and filtering artifacts introduced by problematic upsampling operators, and (ii) the spectral replicas that emerge while upsampling. We then compare different upsampling layers, showing that nearest neighbor upsamplers can be an alternative to the problematic (but state-of-the-art) transposed and subpixel convolutions which are prone to introduce tonal artifacts.

Random Filters for Compressive Sampling

Abstract: This paper discusses random filtering, a recently proposed method for directly acquiring a compressed version of a digital signal. The technique is based on convolution of the signal with a fixed FIR filter having random taps, followed by downsampling. Experiments show that random filtering is effective at acquiring sparse and compressible signals. This process has the potential for implementation in analog hardware, and so it may have a role to play in new types of analog/digital converters.

ICIF-Net: Intra-Scale Cross-Interaction and Inter-Scale Feature Fusion Network for Bitemporal Remote Sensing Images Change Detection

Abstract: Change detection (CD) of remote sensing (RS) images has enjoyed remarkable success by virtue of convolutional neural networks (CNNs) with promising discriminative capabilities. However, CNNs lack the capability of modeling long-range dependencies in bitemporal image pairs, resulting in inferior identifiability against the same semantic targets yet with varying features. The recently thriving Transformer, on the contrary, is warranted, for practice, with global receptive fields. To jointly harvest the local-global features and circumvent the misalignment issues caused by step-by-step downsampling operations in traditional backbone networks, we propose an intra-scale cross-interaction and inter-scale feature fusion network (ICIF-Net), explicitly tapping the potential of integrating CNN and Transformer. In particular, the local features and global features, respectively, extracted by CNN and Transformer, are interactively communicated at the same spatial resolution using a linearized Conv Attention module, which motivates the counterpart to glimpse the representation of another branch while preserving its own features. In addition, with the introduction of two attention-based inter-scale fusion schemes, including mask-based aggregation and spatial alignment (SA), information integration is enforced at different resolutions. Finally, the integrated features are fed into a conventional change prediction head to generate the output. Extensive experiments conducted on four CD datasets of bitemporal (RS) images demonstrate that our ICIF-Net surpasses the other state-of-the-art (SOTA) approaches.

Multi-rate digital phase locked loop

Abstract: A Digital Phase-Locked Loop (DPLL) involves a Time-to-Digital Converter (TDC) that receives a DCO output signal and a reference clock and outputs a first stream of digital values. Quantization noise is reduced by clocking the TDC at a high rate. Downsampling circuitry converts the first stream into a second stream. The second stream is supplied to a phase detecting summer of the DPLL such that a control portion of the DPLL can switch at a lower rate to reduce power consumption. The DPLL is therefore referred to as a multi-rate DPLL. A third stream of digital tuning words output by the control portion is upsampled before being supplied to the DCO so that the DCO can be clocked at the higher rate, thereby reducing digital images. In a receiver application, no upsampling is performed and the DCO is clocked at the lower rate, thereby further reducing power consumption.

Directional Equalization of Sparse Head-Related Transfer Function Sets for Spatial Upsampling

Abstract: Acquiring decent full-spherical sets of head-related transfer functions HRTFs based on a small number of measurements is highly desirable. For spatial upsampling, HRTF interpolation in the spatially continuous spherical harmonics SH domain is a common approach. However, the number of measured HRTFs limits the assessable SH order, resulting in order-limited HRTFs when transformed to the SH domain. Thus, the SH representation of sparse HRTF sets shows restricted spatial resolution and suffers from order-limitation errors. We present a method that reduces these errors by a directional equalization prior to the SH transform. This is done by a spectral division of each HRTF with a corresponding directional rigid sphere transfer function. The processing removes direction-dependent temporal and spectral components and, therefore, significantly reduces the spatial complexity of the HRTF set, allowing for an enhanced interpolation of HRTFs at reduced SH orders. Spatial upsampling is achieved by an inverse SH transform on an arbitrary dense sampling grid. A subsequent de-equalization by a spectral multiplication with the rigid sphere transfer function recovers the energy in higher spatial orders that was not inherent in the sparse HRTF set. For evaluation, HRTFs were calculated for various limited orders from sparse datasets and compared to a reference. The results show that the proposed method clearly outperforms common SH interpolation of HRTF spectra regarding the overall spectral and temporal structure as well as modeled localization performance.