Software
"All codes are copyrighted by PI Keigo Hirakawa. The softwares are for research use only. Use of software for commercial purposes without prior agreement with the authors is strictly prohibited. We do not guarantee the code's accuracy. Patent applications have been filed for many of these algorithms. We would appreciate if acknowledgments were made for the use of our codes in your publications."
Under construction. We will be adding more software packages soon.
Adaptive Homogeneity-Directed (AHD) Demosaicking
AHD Demosaicking is the default algorithm used in DCRAW.
Keigo Hirakawa, and Thomas W. Parks. "Adaptive homogeneity-directed demosaicing algorithm." IEEE Transactions on Image Processing 14, no. 3 (2005): 360-369.
CCFind automatically detects Macbeth ColorChecker inside an image
Stochastic Bilateral Filter (SBF) / Stochastic Non-Local Means (SNLM)
Accelerated bilateral filter/non-local means implementation using Monte-Carlo.
SBF is invariant to the window size, edge image color dimension. SNLM is also invariant to the block size. FSBF is also invariant to filter image color dimension.
Christina Karam, and Keigo Hirakawa. "Monte-Carlo acceleration of bilateral filter and non-local means." IEEE Transactions on Image Processing 27.3 (2017): 1462-1474.
Christina Karam, Kenjiro Sugimoto, and Keigo Hirakawa. "Near-Constant Time Bilateral Filter for High Dimensional Images." In 2018 25th IEEE International Conference on Image Processing (ICIP), pp. 3244-3248. IEEE, 2018.
Christina Karam, Chong Chen, and Keigo Hirakawa. "Stochastic bilateral filter for high-dimensional images." In 2015 IEEE International Conference on Image Processing (ICIP), pp. 192-196. IEEE, 2015.
SBF/SNLM [Reference Code] (fast!)
FSBF [Reference Code] (faster!)
Fast Convolutional Distance Transform
Accelerated distance transform using soft-maximum
The proposed acceleration technique is "distance metric-agnostic."
Complexity is O(N log N). It reduces to O(N) in Euclidean norm.
Christina Karam, Kenjiro Sugimoto, and Keigo Hirakawa. "Fast Convolutional Distance Transform." IEEE Signal Processing Letters 26.6 (2019): 853-857.
CAMRA (Camera-Aware Multi-Resolution Analysis) For RAW Image Sensor Compression
Wavelet-based compression algorithm for Bayer raw sensor data.
CAMRA-D is a lossy compression algorithm. CAMRA-A is a lossless compression.
Yeejin Lee, Keigo Hirakawa, and Truong Q. Nguyen. "Camera-aware multi-resolution analysis for raw image sensor data compression." IEEE Transactions on Image Processing 27, no. 6 (2018): 2806-2817.
Yeejin Lee, Keigo Hirakawa, and Truong Q. Nguyen. "Lossless compression of CFA sampled image using decorrelated Mallat wavelet packet decomposition." In 2017 IEEE International Conference on Image Processing (ICIP), pp. 2721-2725. IEEE, 2017.
CAMRA-D [Reference Code] (lossy compression)
CAMRA-A [Reference Code] (lossless compression, coming soon)
Combining Inertial Measurements with Blind Image Deblurring Using Distance Transform
Blind image deblurring with the help of IMU. We estimate the PSF caused by camera motion using *both* IMU and blurred image. Works better than IMU-only deblurring and blurred image-only deblurring.
Yi Zhang, and Keigo Hirakawa. "Combining inertial measurements with blind image deblurring using distance transform." IEEE Transactions on Computational Imaging 2, no. 3 (2016): 281-293.
Joint Demosaicking and Denoising
Posterior Sparsity-Directed Demosaicking (PSDD)
Wavelet-based demosaicking that can be combined with any wavelet-based denoising method.
PSDD is also a chromatic aberration-tolerant demosaicking method
Jiachao Zhang, Andong Sheng, and Keigo Hirakawa. "A Wavelet-GSM Approach to Demosaicking." IEEE Signal Processing Letters 25, no. 6 (2018): 778-782.
Jan Tore Korneliussen, and Keigo Hirakawa. "Camera processing with chromatic aberration." IEEE Transactions on Image Processing 23, no. 10 (2014): 4539-4552.
Keigo Hirakawa, Xiao-Li Meng, and Patrick J. Wolfe. "A framework for wavelet-based analysis and processing of color filter array images with applications to denoising and demosaicing." In 2007 IEEE International Conference on Acoustics, Speech and Signal Processing-ICASSP'07, vol. 1, pp. I-597. IEEE, 2007.
Total Least Square Demosaicking (TLSD)
Pixel domain demosaicking that is an extension of the total least squares denoising.
Keigo Hirakawa, and Thomas W. Parks. "Joint demosaicing and denoising." IEEE Transactions on Image Processing 15, no. 8 (2006): 2146-2157.
Multiplicative Multiscale Innovation (MMI)
Denoising of image contrast
Wu Cheng, and Keigo Hirakawa. "Towards optimal Denoising of image contrast." IEEE Transactions on Image Processing 27, no. 7 (2018): 3446-3458.
Skellam Mean Estimation
Wavelet-based image denoising based on minimum risk shrinkage operator
Wu Cheng, and Keigo Hirakawa. "Minimum risk wavelet shrinkage operator for poisson image denoising." IEEE Transactions on Image Processing 24, no. 5 (2015): 1660-1671.
S4HDR takes advantage of the differences in the sensitivities of red, green, and blue pixels to recover high dynamic range images.
[tpsingle id="11"]
DDWT based estimation of motion blur + deblurring (CVPR 2013)
DDWT based estimation of defocus blur + deblurring (CVPR 2013)
DDWT based camera shake deblurring (work in progress)
Yi Zhang, and Keigo Hirakawa. "Blur processing using double discrete wavelet transform." In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp. 1091-1098. 2013.
Our color constancy method makes use of statistical model for the spatial distribution of colors in white balanced images to infer illumination parameters as those being most likely under our model.
[tpsingle id="8"]
Although pixel binning significantly improves noise performance of an image sensor, it causes severe aliasing artifacts. Our binning-aware demosaicking method is designed to eliminate these problems.
Xiaodan Jin, and Keigo Hirakawa. "Analysis and processing of pixel binning for color image sensor." EURASIP Journal on Advances in Signal Processing 2012, no. 1 (2012): 125.
Chromatic aberration-tolerant demosaicking (PSDD)
Separate recovery methods for spatially lowpass and highpass image details
Jan Tore Korneliussen, and Keigo Hirakawa. "Camera processing with chromatic aberration." IEEE Transactions on Image Processing 23, no. 10 (2014): 4539-4552.
Neuromorphic Camera Denoising
Inceptive event is a temporal only filter designed to remove noise in event cameras
Baldwin R.W., Almatrafi M., Kaufman J.R., Asari V., Hirakawa K. (2019) Inceptive Event Time-Surfaces for Object Classification Using Neuromorphic Cameras.
Microscopy
L's approach to illumination non-uniformity correction leverages shifted images with overlapping scene content to estimate the illumination profile of optical path.
Coming soon...
Spatial-Spectral Color Filter Array Design
Universal Demosaicking
Poisson noise parameter estimation