Metody zvyšování roslišení obrazů

The methods of image resolution enhancement

conference paper

en

For wider using of industrial CCD cameras as measurement tools in production defend the resolution of this one what closely pertinent to price of these cameras. In cooperation with some grant projects we adapt and tried out some methods for extraction of one high-resolution image from serie of undersampled images for their using in industry. These methods are based on sub-pixel movement of the same images and their interpolation to one image. The method is divided to two parts: 1. The recognition of sub-pixel movement 2. Interpolation of images The recognition of sub-pixel movement The recognition of sub-pixel movement was doing by using of first order of Taylor series. Unfortunately if we are using this method for estimation of shifting we are getting the best result in case that the shifts are close to 0.5 point. If the shifts are bigger the estimations are inaccurate. For this reason we are using Cross-correlation for estimation of integer shifts and after that previous method for sub-pixel shift estimation. Interpolation of images The second part is an image interpolation. If the sub-pixel of images is known the interpolation is made by fitting plane throught points of images.

Širšího nasazení průmyslových CCD videokamer jako měřícího nástroje do výroby v současné době do jisté míry brání rozlišení používaných CCD elementů u těchto kamer, a v závislosti na tomto také jejich cena. V rámci prací na našem ústavu jsme zpracovali metody jež umožňují matematicky zvýšit rozlišovací schopnost a následně i cenu použití v průmyslové výrobě.

For wider using of industrial CCD cameras as measurement tools in production defend the resolution of this one what closely pertinent to price of these cameras. In cooperation with some grant projects we adapt and tried out some methods for extraction of one high-resolution image from serie of undersampled images for their using in industry. These methods are based on sub-pixel movement of the same images and their interpolation to one image. The method is divided to two parts: 1. The recognition of sub-pixel movement 2. Interpolation of images The recognition of sub-pixel movement The recognition of sub-pixel movement was doing by using of first order of Taylor series. Unfortunately if we are using this method for estimation of shifting we are getting the best result in case that the shifts are close to 0.5 point. If the shifts are bigger the estimations are inaccurate. For this reason we are using Cross-correlation for estimation of integer shifts and after that previous method for sub-pixel shift estimation. Interpolation of images The second part is an image interpolation. If the sub-pixel of images is known the interpolation is made by fitting plane throught points of images.

image enhancement, sub-pixel motions, image interpolation

2002

25.10.2000

Vysoká škola báňská, TU Ostrava

Ostrava

80-7078-890-9

Instruments and Control

1