>> Top >> Project >> Main research results >> Vision and image information processing group

Object image (upper), digital hologram (lower left) and reconstructed images (lower right)

Generation of fractal fields of scattered light and their properties.

By devising optical systems for producing light scattering phenomena, it is possible to generate scattered light fields with fractal properties such as self-similarity and power-law correlation. So far, such methods have been proposed theoretically and demonstrated experimentally. In the present study, further considerations were given on the basis of computer simulations and some new knowledge was obtained concerning the conditions of the illumination optics for the scattered fields to have fractality and concerning multifractality of the scattered fields.

Intensity distribution of a fractal field of scattered light (experiment)

Generation of nondiffracting fields of scattered light and their applications

A light beam that keeps its shape and size during free-space propagation is called nondiffracting and a Bessel beam is known as one of most typical nondiffracting beams. It was shown that it is possible to provide scattered fields with such a property, and generated nondiffracting scattered fields were applied to develop a method for measuring in-plane motion of an object with high out-of-plane insensitivity. The figure on the right shows measurement curves for two cases of in-plane motion only and of combined motions of in- and out-of-plane with the latter 100 times larger than the former. Two curves are almost identical, indicating high insensitivity to the out-of-plane motion of the object.

Intensity distributions of a nondiffracting field of scattered light (experimental)

Measurement of motion of an object using nondiffracting scattered fields. Results in the cases of in-plane motion only (broken line) and of combined motions of in- and out-of-plane with the latter 100 times larger than the former(solid line)

Discriminant methods of winter pavement conditions

Winter pavement shows various appearances depending upon its kind and snow conditions. To develop methods for automatically discriminating its conditions from images captured by CCD camera, various statistical analyses are applied to different pavement images. One of the results is shown in the following figure. Six different conditions for two kinds of pavement are roughly discriminated by means of three parameters of skewness and contrast derived from probability density functions and fractal dimension of the pavement images.

Discrimination of winter pavement conditions by means of fractal dimension of the images
(the left and right figures constitute a stereogram).

Near infrared spectral analysis by means of generalized differentiations
Spectra in the near infrared region have properties suitable for analyzing constituents of complex organic objects such as biomedical tissues, agricultural products and foods, and hence analyses on the basis of the near infrared spectroscopy (NIRS) are expanding rapidly into various industries. In this study, to further improve the performance of NIRS, data processing based on generalized differentiations is applied. The following figure shows a result of correlation map between generalized derivative spectra of rice flour and its major constituents (amylose, moisture and protein). As the derivative order increases, correlation peaks (portions of dark red) narrow in width, separate into two or more, and/or emerge.

Correlation spectra between generalized derivative spectra of rice flour and its major constituents (amylose, moisture and protein)

>> Speech and language information processing group
>> Robot and control information processing group