Space-Variant Visual Sensor with Color Acquisition (SVAVISCA)

 ESPRIT Long Term Research Project No. 21951

See our latest Press Release


Project Synopsis

The use of space-variant visual sensors in image communication and processing is gaining more and more attention as a simple and direct way of reducing the visual information transmitted and/or processed while preserving both high resolution and a wide field of view. At present two different implementations of the space-variant concept have been realized at IMEC in collaboration with DIST. The first of such sensors was fabricated using CCD technology (the chip contained about 2,000 photocells) in 1989 The second, using CMOS technology and about 8,000 photocells, has been designed and fabricated within an ongoing EU-supported project (TIDE project IBIDEM P1038) and is now in the testing phase. Both sensors acquire black and white images.

However the possibility of acquiring color images, is bound to become an important factor particularly for some image processing applications (e.g. quality control and surveillance) and for image transmission (e.g. multimedia communication, videotelephony, etc.).

The aim of this project is to demonstrate the feasibility of a space-variant silicon sensor with color acquisition capability.

To achieve this goal we intend to investigate two different technological approaches:

The two solutions have, of course, advantages and disadvantages. The integrated solution is, in perspective, the most interesting one. Among the pros: i) cost will be lower, ii) the system will be simpler, iii) the technology for color chip fabrication (particularly for special purpose applications) is strategically important for the European industry. On the converse, the deposition of color microfilters, reduces the resolution of the overall image and, therefore, requires, a comparatively higher number of pixels.

The "system" solution has the advantage of exploiting all the spatial resolution available on the chip at the expenses of a higher cost and complexity. The reduced temporal resolution (required to acquire the sequence of RGB components) is not really a problem because, due to the relatively small number of pixels acquired by a space-variant sensor, the "absolute" frame rate can be higher than conventional cameras (with the same technology) providing, at the end, the required frame rate.

The first phase of the project is aimed at building two prototypes adopting the two solutions described above. These solutions will then be compared and, according to the overall results and to the results of the comparison, a second phase project will be submitted, if necessary, to design a fully integrated space-variant color camera.

The assessment of the results of the project will be based on the experience acquired and the procedures used in the IBIDEM project where the user's acceptability of the proposed solution was tested experimentally with the help of hard-of-hearing and deaf persons. Qualitative as well as quantitative data will be collected and comparison with the already available data based on a Black and White space-variant solution will be reported.

Contact Person

Prof. Giulio Sandini		tel. +39-10-353.2779
DIST - University of Genova	fax +39-10-353.2154
Via Opera Pia, 13		e-mail: giulio@dist.unige.it
16145 Genova - Italy

Consortium

Participants		Role (Nation)	Principal Investigators
DIST			Coordinator (I) Giulio Sandini, Paolo Questa
IMEC			Partner (B)	Lou Hermans, Bart Dierickx, Danny Scheffer
UNITEK Consortium	Partner (I)	Fabrizio Ferrari, Andrea Mannucci, Francesco Buemi