Objects recognition based on their image

 

Essence

It is all known, that objects image analysis allow getting maximum information about it. One of the important directions of using automatic objects recognition system is image recognition, got in visible light under natural lightening.

There are two methods of solving a recognition invariance problem: invariant attribute vector ensemble forming and special classificatory functioning procedures using, which provide invariant recognition of nonnormal attribute vector.

Attribute vector, which is invariant to objects image affine transformation, can be formed using this image shape description. Radiation propagation through turbulent atmosphere lead to imaging system angular resolution limitation, located objects image blur and, in result, image shape distortion. We have developed image modeling software of objects that is observed through turbulence atmosphere with image scattering, to form a representative ensemble of attribute vector, which adequate represent such situations.
 

Scattering function modeling results counted for point and objects image,
observed through turbulence atmosphere

 

Using of active contour modified method allow to construct objects image closed loop in the presence of additive noise.

 

 Initial real objects image and chosen shapes of image realizations
at different additive noise dispersion 

 

Neural network of direct propagation, taught on attribute vector ensemble, ensure stable objects recognition by their images even at viewing conditions changes and image distortion by turbulence atmosphere.
 

References

  • Бурый Е.В., Джафар А., "Распознавание двухмерных контрастных изображений объектов по инвариантным информативным признакам с применением нейросетевых алгоритмов" // Нейрокомпьютеры: разработка, применение. – 2001. – № 1.
  • Buryi E., Jafar A. Invariant Features Extraction Forming for Contrast Planer Objects Recognition // 8–th Int. Conf. ACS’2001. – Szczecin ( Poland), 2001.– Part . 2.
  • Бурый Е.В., Джафар А., "Получение инвариантных информативных признаков для распознавания двухмерных контрастных изображений объектов" // Нейрокомпьютеры и их применение: Труды VI Всеросс. конф. – М., 2000.
  • Аракчеев П.В., "Применение нейросетевых алгоритмов для распознавания контрастных телевизионных изображений" // Международный форум информатизации: Доклады международной конференции «Информационные средства и технологии» – М.: Изд.-во «Станкин», 2001. – Т. 1.