Fabrizio Ciancetta, Giovanni Bucci and Edoardo Fiorucci
University of L'Aquila

In industrial society, the artificial vision has an important role in assembly-line to control the quality of the process. These controls are demanded to measurement system that estimates the goodness product measuring the parameters process imposed by factory. Different methods are used in order to control the final product adopting human approach with specific labour or automatic system with sensors or video cameras. While the automatic system based on sensors can be used on product with limited parameters to control, the automatic system based on video cameras gives a global vision of product acquiring only a single frame. The cost of the automatic system is very low due to wide diffusion of CCD and CMOS chip. The image processing and the development of algorithm to process image have an important role in this area. It has been developed a DSP based implementation of an algorithm dedicates to pre-process images from a video camera. During the entire process-line, the product presents many manipulation and transformation changing its position and its shape. The algorithm developed rotates the acquired image to send successively to the process control system, which compare it with a reference image. The processing has been developed using a Texas Instruments DSP to reduce the processing time.