WIRE: Wavelet Implicit Neural Representations
WIRE: Wavelet Implicit Neural Representations
复制标题
DOI：
10.1109/cvpr52729.2023.01775
复制
发表时间：
2023-01
期刊：
影响因子：
--
通讯作者：
Vishwanath Saragadam;Daniel LeJeune;Jasper Tan;Guha Balakrishnan;A. Veeraraghavan;Richard Baraniuk
中科院分区：
文献类型：
--
作者：
Vishwanath Saragadam;Daniel LeJeune;Jasper Tan;Guha Balakrishnan;A. Veeraraghavan;Richard Baraniuk
内隐神经表征（INRs）近年来在许多视觉相关领域取得了进展。INR性能在很大程度上取决于其MLP网络中使用的激活函数的选择。已经探索了广泛的非线性，但不幸的是，目前设计具有高精度的inr也具有较差的鲁棒性（对信号噪声，参数变化等）。受谐波分析的启发，我们开发了一种新的，高度准确和强大的INR，不会表现出这种权衡。我们的小波隐式神经表示（WIRE）使用复杂Gabor小波作为其激活函数，该小波以最佳地集中在空间频率上而闻名，并且在表示图像时具有出色的偏差。广泛的实验（图像去噪、图像上漆、超分辨率、计算机断层扫描重建、图像过拟合和神经辐射场的新视图合成）表明，WIRE在INR精度、训练时间和鲁棒性方面定义了新的技术状态。
Implicit neural representations (INRs) have recently advanced numerous vision-related areas. INR performance depends strongly on the choice of activation function employed in its MLP network. A wide range of nonlinearities have been explored, but, unfortunately, current INRs designed to have high accuracy also suffer from poor robustness (to signal noise, parameter variation, etc.). Inspired by harmonic analysis, we develop a new, highly accurate and robust INR that does not exhibit this trade off. Our Wavelet Implicit neural REpresentation (WIRE) uses as its activation function the complex Gabor wavelet that is well-known to be optimally concentrated in space-frequency and to have excellent biases for representing images. A wide range of experiments (image denoising, image inpainting, super-resolution, computed tomography reconstruction, image over fitting, and novel view synthesis with neural radiance fields) demonstrate that WIRE defines the new state of the art in INR accuracy, training time, and robustness.