Total variation models based algorithm of illumination normalization for face recognition

Yan, Kai; Ren, Hongliang; Yu, Hailong

doi:10.1109/cisp.2010.5647075

Cited by 2 publications

(1 citation statement)

References 6 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This algorithm can be applied to various computer vision applications. For example, the Total Variation algorithm can be used to preprocess face recognition algorithms to improve the result [1]. The algorithm is used optical flow processing [2], and medical image reconstruction [3].…”

Section: Introductionmentioning

confidence: 99%

Design and implementation of hardware architecture for denoising using FPGA

Jeon

Lee

Jin

et al. 2013

2013 IEEE 9th International Colloquium on Signal Processing and Its Applications

View full text Add to dashboard Cite

Noise removal in image processing is required in a variety of fields such as object tracking, stereo vision and medical image reconstruction. To obtain accurate results, various video pre-processing is required. We propose a hardware architecture using FPGA to improve the processing speed with the Total Variation algorithm for noise removing images. In the proposed system, we can process images with a resolution of 640ⅹ 480. We remove noise from the input noisy image, after 10 cycles of operations. In the first step, we obtain the right, bottom and center pixel values and the differences to obtain. In the second step, we add pixels to the center of the operation parameters, and the difference between the values are obtained central pixel. The operation parameters and the difference of the values of the surrounding pixels are reflected in the following operations. We repeat this process 10 times to remove the noise in the image. The noise removal performance is better than prior results, but the operation is complex and requires considerable computing power. We implemented the proposed system in hardware that requires high computing power for real-time processing with these processes. The processing delay is 0.8ms. We designed pipeline architecture to delay the operation. The proposed system can operate on image resolution of 640 ⅹ480 with a speed of 250Mhz.

show abstract

Section: Introductionmentioning

confidence: 99%