## Reconfigurable computing with Field

### BibTeX

@MISC{Kruzner_reconfigurablecomputing,

author = {Andrew Kruzner and Jack Merwin and Derek Rollend and Ananth Nallamuthu and Chirag Gupta and Melissa C. Smith},

title = {Reconfigurable computing with Field},

year = {}

}

### OpenURL

### Abstract

### Citations

3030 | Eigenfaces for recognition
- Turk, Pentland
- 1991
(Show Context)
Citation Context ...odels, and the Bayesian Intrapersonal/ Extrapersonal Classifier (BIC) [10-12]. One of the oldest and most efficient algorithms for performing face recognition is the PCA based Eigen face construction =-=[13]-=-. In this algorithm the test images are stored in the form of Eigenvectors. Each Eigenvector corresponds to a face image. Faces are represented by their projection onto a subset of Eigenvectors. So ea... |

225 | Face recognition by independent component analysis
- Bartlett, Movellan, et al.
- 2002
(Show Context)
Citation Context ... be implemented using two architectures viz ICA1 and ICA2. ICA1 treats the images as random variables where pixels are the outcome; ICA2 treats pixels as random variables where images are the outcome =-=[14]-=-. Finally, the LDA-based face recognition is similar to PCA, but in LDA the Eigenvalues are calculated using a separation matrix rather than a covariance matrix [15]. The benefits of implementing thes... |

210 | Discriminant analysis for recognition of human face images
- Etemed, Chellappa
- 1997
(Show Context)
Citation Context ...ables where images are the outcome [14]. Finally, the LDA-based face recognition is similar to PCA, but in LDA the Eigenvalues are calculated using a separation matrix rather than a covariance matrix =-=[15]-=-. The benefits of implementing these compute intensive algorithms with FPGA-based systems include the potential power, space, and cost savings vs. traditional CPUs. As mentioned earlier, FPGAs are als... |

40 | Pipeline Vectorization for Reconfigurable Systems
- Weinhardt, Luk
- 1999
(Show Context)
Citation Context ...s working on the same data) can be implemented to run in parallel on the FPGA as demonstrated by several large-scale data manipulation applications such as encryption [8], and string pattern matching =-=[9]-=-. The MISD capability of FPGAs will be exploited to process the same input data with a heterogeneous set of algorithms simultaneously. This capability will enable real-time comparison of algorithm res... |

24 |
Independent comparative study
- Delac, Grgic, et al.
(Show Context)
Citation Context ...ify matches even when the compared images differ in such attributes. Though a number of facial recognition algorithms are available, none of them is robust enough for use with all types of input data =-=[5]-=-. A consensus made by comparing the results of multiple algorithms could improve the reliability of the facial recognition system. But processing data through multiple algorithms sequentially would be... |

24 | Analysis of pca-based and fisher discriminant-based image recognition algorithms
- Yambor
- 2000
(Show Context)
Citation Context ..., L2 norm, covariance, Mahalanobis distance, and correlation created by the Eigenvectors of the covariance matrix of the training data and the subspace created by the Fisher basis vectors of the data =-=[2]-=-. The final decision is made using a classifying algorithm like Bayesian classification, AdaBoost or Support Vector Machine [3]. The selection of a specific subspace depends greatly on the situation w... |

21 |
Using FPGA devices to accelerate biomolecular simulations
- Alam, Agarwal, et al.
- 2007
(Show Context)
Citation Context ...tiple algorithms can be implemented in FPGA during different segments of the runtime thereby supporting the overall application. FPGAs have previously been used to accelerate biomolecular simulations =-=[7]-=- by implementing computational bottlenecks in the reconfigurable hardware. Alternatively, several algorithms (either homogeneous copies of a single algorithm working on different data or heterogeneous... |

13 | An fpga implementation and performance evaluation of the serpent block cipher
- Elbirt, Paar
(Show Context)
Citation Context ...t data or heterogeneous algorithms working on the same data) can be implemented to run in parallel on the FPGA as demonstrated by several large-scale data manipulation applications such as encryption =-=[8]-=-, and string pattern matching [9]. The MISD capability of FPGAs will be exploited to process the same input data with a heterogeneous set of algorithms simultaneously. This capability will enable real... |

10 |
Biometric Recognition: How Do I Know Who You Are
- Jain
- 2004
(Show Context)
Citation Context ...suited for applications where multiple algorithms utilized the same data set in parallel. 2. Background Table 1 compares the most commonly used features for identification: fingerprint, iris and face =-=[1]-=-. The low, medium, and high ratings represent the applicability, robustness, reliability and performance of the different biometric techniques. For example, biometric identification via facial recogni... |

10 |
Facial similarity across age, disguise, illumination and pose
- Ramanathan
- 1999
(Show Context)
Citation Context ...to variations in illumination conditions, facial hair, facial expressions, temporal changes, ageing factors in the face, etc., leading to increased computing complexity and lower recognition accuracy =-=[4]-=-. The facial recognition algorithm used should be able to identify matches even when the compared images differ in such attributes. Though a number of facial recognition algorithms are available, none... |

10 | The bayesian intrapersonal/extrapersonal classifier - Teixeira - 2003 |

9 | Data characteristics that determine classifier performance
- Walt, Barnard
(Show Context)
Citation Context ...ng data and the subspace created by the Fisher basis vectors of the data [2]. The final decision is made using a classifying algorithm like Bayesian classification, AdaBoost or Support Vector Machine =-=[3]-=-. The selection of a specific subspace depends greatly on the situation where the recognition is to be done. Furthermore, variations within the subspace also effect performance. For example, the selec... |

3 |
HighPerformance Reconfigurable Computing Application Programming
- Kindratenko, Pointer, et al.
- 2006
(Show Context)
Citation Context ... for embedded applications and in real-time (or pseudo) performance for timely identification. FPGA-based systems are emerging as an alternative to microprocessors for several classes of applications =-=[6]-=-. Their lower clock speeds keep power consumption low and performance is maintained with application specific hardware implementation. Via runtime reconfiguration, multiple algorithms can be implement... |

1 | Elastic Bunch Graph Mapping - Bolme - 2003 |

1 | EBGM vs subspace projection for face recognition - Stergiou, Pnevmatikakis, et al. |

1 |
Robust Preprocessing Algorithm for Face Recognition
- Mittal, Sasi
(Show Context)
Citation Context ...with a recognition algorithm, the image must be pre-processed into its mathematical representation. Preprocessing may also include image normalization steps to eliminate variations due to facial hair =-=[16]-=-, illumination, posture, expression, scale etc., and thereby aid the recognition process. In a typical recognition process, the training images and candidate image are projected onto a subspace such a... |