Modified composite filter for pattern recognition in the presence of noise with a non-zero mean

Arsenault, Henri H.; Sheng, Yunlong; Bulabois, J.

doi:10.1016/0030-4018(87)90214-8

Cited by 17 publications

(4 citation statements)

References 6 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…From previous work done by Arsenault et al 5 it is possible to obtain a correlation plane invariant against a constant additive factor. This can be accomplished by means of an synthetic discriminant function filter h͑x͒ that recognizes the object f ͑x͒, but that discriminates against the object support ◊͑x͒.…”

Section: Intensity Invariant Recognitionmentioning

confidence: 99%

“…Kotynski and Chalasinska 3,4 generalized this algorithm by using the Hölder inequality, and ended up with a whole new family of solutions of which the latter is a special case. Arsenault et al 5 proved that it is possible to achieve additive intensity invariance using only one correlation. They used a composite 6 or synthetic discriminant filter 7 constructed to recognize the target but to discriminate against the target support ͑a binary image equal to unity, where the target is present and equal to zero everywhere else͒.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Recognition of unsegmented targets invariant under transformations of intensity

et al. 2002

Self Cite

View full text Add to dashboard Cite

Images taken in noncooperative environments do not always have targets under the same illumination conditions. There is a need for methods to detect targets independently of the illumination. We propose a technique that yields correlation peaks that are invariant under a linear intensity transformation of object intensity. The new locally adaptive contrast-invariant filter accomplishes this by combining three correlations in a nonlinear way. This method is not only intensity invariant but also has good discrimination and resistance to noise. We present simulation results for various intensity transformations with and without random and correlated noise. When the noise is high enough to threaten errors, the method trades off intensity invariance in order to achieve the optimum signal to noise ratio, and the peak to sidelobe ratio in the presence of clutter is always greater than one. In the presence of random disjoint noise, the signal to noise ratio is independent of the target contrast and of the level of the noise.

show abstract

Section: Intensity Invariant Recognitionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Recognition of unsegmented targets invariant under transformations of intensity

et al. 2002

Self Cite

View full text Add to dashboard Cite

show abstract

“…Arsenault et al 2 proved that it is possible to achieve additive intensity invariance using only one correlation. Arsenault and Lefebvre 3 used the homomorphic transformation to transform the multiplicative intensity problem to an additive intensity problem.…”

Section: Introductionmentioning

confidence: 98%

Recognition in the presence of nonuniform illumination

Lefebvre¹,

Arsenault²,

Roy³

2003

Optical Information Systems

Self Cite

View full text Add to dashboard Cite

Linear filtering methods often fail to detect dark objects because the correlation peak height is proportional to the object intensity. The strong response from very bright clutter can result in false detections. In addition, intensity variations of targets occur in real world situations. We address the problem of image recognition in the presence of unknown uniform and nonuniform intensity transformations. The uniform intensity transformations considered are multiplicative and additive. This part of the problem has already been solved using a method that we recently proposed called the LACIF (Locally Adaptive Contrast-Invariant Filter). In this paper we generalize this method for situations where a linear intensity gradient across an object can be present. We use a set of four orthonormal images, and compute the correlations between each of those images and the scene. We carry out one additional correlation and combine the five correlation planes in a nonlinear manner. Results show that discrimination is good.

show abstract

“…Traditionally, zero calibrations and alignments are used to remove non-zero-mean noise from a measurement. Parameters of non-zero-mean noise are estimated in order to eliminate this kind of noise [2]. Usually, a zero calibration is performed when measuring systems are stationary and dynamic smoothing techniques are implemented by various filters [3].…”

Section: Introductionmentioning

confidence: 99%

Unbiasedness of simultaneous independent measurement

Yang¹,

Wang²,

Li³

2012

Meas. Sci. Technol.

View full text Add to dashboard Cite

We manage to get the true parameter value out of biased data without any zero calibration, a result which has not been reported to our knowledge. In order to obtain the true value of a given physical quantity out of biased noises, a general model and an algorithm based on simultaneous independent measurements are demonstrated. A decision function is constructed using variable measuring vectors. A nonlinear estimator is provided using linear combinations of differential statistics of initial observations. Repeated gyroscope experiments targeting the Earth's local rotation velocity are implemented to validate the algorithm and the estimator. Experimental results show that the true rotation velocity is obtained even if initial observations are either positively or negatively biased. Therefore, it is possible to get the true value of a physical quantity using simultaneous multi-dimensional measurements, no matter whether systematic biases exist or not. This reveals a phenomenon that physical observations, even though obscured by time-varying biases and noises, have an intrinsic true-value feature which has not been made use of in current theories and applications.

show abstract

Modified composite filter for pattern recognition in the presence of noise with a non-zero mean

Cited by 17 publications

References 6 publications

Recognition of unsegmented targets invariant under transformations of intensity

Recognition of unsegmented targets invariant under transformations of intensity

Recognition in the presence of nonuniform illumination

Unbiasedness of simultaneous independent measurement

Contact Info

Product

Resources

About