TY - GEN
T1 - Algorithm study of infrared image MMSE filtering
AU - Ping, Qingwei
PY - 2010
Y1 - 2010
N2 - The conventional algorithm of the image filtering is basis on the assumption that the image is stationary. The algorithm based on this model can reduce the noise in the image, but it can also lose the high frequency information. Therefore, the model of the image is improved, so that effect of the image filtering algorithm is improved. This paper assumes that the model of image is the local stationary Gauss model by the analysis of the original infrared image. Furthermore, this paper thinks the noise of the infrared image is not additive noise, but is multiplicative noise. It is the signal-dependent noise. Therefore, the infrared image filtering algorithm based on the minimum mean-square error estimation is devised. Final, this algorithm is compared with the infrared image filtering algorithm based on the maximum likelihood estimation. This algorithm can not only reduce the noise of the infrared image but also reserve the high frequency information. Especially, the algorithm does not lose the point target in the infrared image.
AB - The conventional algorithm of the image filtering is basis on the assumption that the image is stationary. The algorithm based on this model can reduce the noise in the image, but it can also lose the high frequency information. Therefore, the model of the image is improved, so that effect of the image filtering algorithm is improved. This paper assumes that the model of image is the local stationary Gauss model by the analysis of the original infrared image. Furthermore, this paper thinks the noise of the infrared image is not additive noise, but is multiplicative noise. It is the signal-dependent noise. Therefore, the infrared image filtering algorithm based on the minimum mean-square error estimation is devised. Final, this algorithm is compared with the infrared image filtering algorithm based on the maximum likelihood estimation. This algorithm can not only reduce the noise of the infrared image but also reserve the high frequency information. Especially, the algorithm does not lose the point target in the infrared image.
KW - Minimum mean-square error estimation
KW - The infrared image
KW - The multiplicative noise
UR - http://www.scopus.com/inward/record.url?scp=79955778480&partnerID=8YFLogxK
U2 - 10.1109/ISDEA.2010.101
DO - 10.1109/ISDEA.2010.101
M3 - Conference contribution
AN - SCOPUS:79955778480
SN - 9780769542126
T3 - Proceedings - 2010 International Conference on Intelligent System Design and Engineering Application, ISDEA 2010
SP - 842
EP - 845
BT - Proceedings - 2010 International Conference on Intelligent System Design and Engineering Application, ISDEA 2010
PB - IEEE Computer Society
ER -
