Avalanche multiplication and excess noise have been measured on a series of Al Ga1 As–GaAs and GaAs–Al Ga1 As ( = 0 3 0 45, and 0 6) single heterojunction p-i-n diodes. In some devices excess noise is lower than in equivalent homojunction devices with avalanche regions composed of either of the constituent materials, the heterojunction with = 0 3 showing the greatest improvement. Excess noise deteriorates with higher values of because of the associated increase in hole ionization in the Al Ga1 As layer. It also depends critically upon the carrier injection conditions and Monte Carlo simulations show that this dependence results from the variation in the degree of noisy feedback processes on the position of the injected carriers.

The feasibility of using a well-known twin ER clutch linear reversing mechanism as a robotic actuator is demonstrated. High speed of response, displacement and positional accuracy can be obtained bi-directionally. A validated mathematical model of the apparatus provides a basis for the control strategy. Positional accuracy is enhanced by an ER brake which works in sequence with the clutch excitation switches. The robot arm, in rotational displacement is tested over a large number of cycles, speeds, displacements and inertial loads and the quality of control compared to that of competitive conventional servo motors.

This paper desribes a process to classify vehicles into simple groups in real time, using television pictures as an input source. The image processing is done using rapac, a special purpose machine developed at the University of Sheffield. The high speed algorithm used is called vehicle outline template matching (VOTM) algorithm. This compares the outlines of moving vehicles to a set of template outlines which represent each of the vehicle classes to be recognised. (these templates are only valid over defined regions of the images due to perspective effects.) the method whereby moving vehicle outlines (mo's) are generated is described as is the use of local neighbour operators to clean up the image obtained, the method of generating the class template outline, and the means of comparing the template outline to the moving outline. Problems encountered due to perspective effects when aligning the Mo with the template outline and the methods used to overcome them are also described. Post spatial correlation filtering methods used to increase the discrimination of the recognition process are also described and evaluated. Time domain normalisation is used to compensate for vehicle speed, and moving outline capture and vehicle path are also compensated for. Test results indicate that if the result is positive, then a correct match has been achieved. For the covering abstract of the conference see IRRD 819738. (TRRL)

Describes a process to classify vehicles into simple groups, in real time, using TV pictures as an input source. The image processing is done using RAPAC, a special purpose image processing machine developed at the University of Sheffield

The paper describes the design and operation of a real-time image processing system and outlines one of its application areas. The system consists of a dedicated hardware processor called RAPAC (a reconfigurable attached processor architecture for convolution) and a host computer which is used for algorithm development and RAPAC control. RAPAC uses hardware processor units and multiple image memories, in a software controlled architecture, to process 5 MHz streams of pixel data. This processing rate allows it to process a 256 × 256 pixel image in 20 ms, one field time of a standard TV camera. The result is either a new 256 × 256 pixel image generated from the old image or a reduced data set which describes attributes of features in the image. These attributes are used by the host computer to calculate a decision output concerning the content of the image.

Glass furnaces such as the one shown schematically in Fig. 1 are built with blockwork made of a dense fusion cast refractory such as AZS or Sillimanite. After commissioning they are run continuously for typically 3 years to 6 years before wall and floor erosion causes them to become dangerous at which point they rnust be demolished and rebuilt. The cost is typically £2m so it is important to be able to measure wall thickness in order to obtain maximum working life consistent with safety.