Search

Biological vision processes are at the basis of many studies in the image-processing field. In this context, neural networks developed by S. Grossberg constitute an interesting approach. The paper presents a novel method for the automatic analysis of scatter plots from CHIMERA experimental data based on Grossberg's pre-attentive neural systems. The design and implementation of a system developed for this purpose are illustrated. The proposed method yields satisfactory results also in very noisy cases.

In nuclear physics at intermediate energies, the use of multidetector arrays consisting of thousands of detectors is customary. This paper presents a novel approach to the automatic classification of data collected by the 4/spl pi/ detector CHIMERA, constituted by 1192 telescopes of a thin silicon transparent detector coupled to a thick CsI(Tl) scintillator. The method is based on spatial density data processing and produces frequency distributions of the most significant physical parameters. These one-dimensional spectra can be more easily analyzed than the standard two-dimensional scatter plots. With respect to manual or semiautomatic procedures, presently used for data analysis, our method offers the advantages of drastic time reduction and unbiased criteria for cluster identification. The proposed method can be successfully used in both off-line data analysis and multidetector stability control.

A commercial two-DSP-based board is the basic unit of the on-line computational and control system under development for the multidetector array CHIMERA. The system design includes a small scale parallel and distributed architecture of PCs and DSP boards. The PCs are used both as host computer and interface between user and network. Each PC can host up to three DSP boards. The system is able to accomplish the on-beam control of CHIMERA, by means of the on-line computation of special algorithms, instead of using the usual techniques of special pulses to be sent off-beam. The physical data, collected and organized in UDP packets by the data acquisition system, are read through a 10 Mbit/s Ethernet network. The PC software, running under Windows NT 4.0, is based on threads and assures the communication and program managing as well as the result presentation. The DSP-software is written in C++ and takes care of the communication with the PC and of the computation of algorithms. The preliminary on beam results for the first three wheels (about 100 telescopes), obtained by employing two biprocessor boards located in the PCI bus of a Pentium MMX PC, are also presented.

The installation of CHIMERA multidetector, designed in order to study central collisions in heavy ion reactions at intermediate energy, is going on at LNS and the first experiment with the forward part (688 telescopes) is running since May 1999. The aim of this contribution is to present the status of the project.

A commercial DSP-based board installed in a host-PC was employed for the fast, on-line and real-time computation of special algorithms, in order to perform event selection and operate as a 2nd level trigger. Moreover an ad hoc built interface, realized using PLDs with a view to connecting the DSP-board to the ADCs and to the data acquisition system, has been tested in order to evaluate the performances of these programmable devices used as a look-up-table and as a decisional part of a 1st level trigger.

In nuclear physics at intermediate energies the use of multidetector arrays consisting of thousands of detectors is customary. For each event up to one hundred of detection cells are hit. As usual in the case of a large number of detectors tools are needed for their stability control and the data analysis automation. This paper presents a simple, novel approach to the classification of data collected using a multidetector array. We describe the results obtained by applying special algorithms to the signals coming from the detection cells of the 4/spl pi/ multidetector array CHIMERA, constituted by telescopes of a thin silicon transparent detector coupled to a thick CsI(Tl) scintillator. The method, which does not require any preliminary energy calibration, consists in a quasiautomatic data processing and produces frequency distributions of the most significant physical parameters. These monodimensional spectra are more easily automatically analyzed than the standard bidimensional scatter plots. In particular, in the proposed system the interactive phase is shorter than in the usual analysis, thus saving time and operator intervention. The proposed method has been successfully used both in the off-line data analysis and in the control of the multidetector stability.