Image processing technologies are broadly applied on modern machine vision and industrial inspection , but there is usually a trade-off between accuracy and speed of inspection . We plan to solve the plight by two steps : 1.We will develop many major image processing methods such as image boundary¡Bto remove the noise¡Bpattern match , and so on . 2.We will focus on sub-pixel algorithm and boundary research to improve the image accuracy and processing time by software under limited hardware . As we know , pixel is the most basic element of an image , but we can divide one pixel into several smaller parts by mathematics ; in the meanwhile , the pixel accuracy can be improved . We will use algorithm to realize the goal in continuous way , and research on the flow of image recognition to find out a best flow for any specific image properties .

Flexible printed circuits (FPCs) have a flexible character, so the topic for high accuracy and speed of drill is important. We will create an automatic system that joins vision-guided function to accomplish the challenge object of high quality and low cost. The system must contain two sub-systems, which are the machine position control and the image recognition. The machine position control system basis framework moves to position after getting hole-position with different methods and scheme of trajectory planning. The image recognition system framework exports correction to machine position control system that integrates the technique of charge-coupled device (CCD), light source design, snap an image in region of interest (ROI) with image grabber card, pattern match that uses normalized cross correlation (NCC) algorithm. We can proof that the system can achieve the expected goal of high speed and accuracy of drill.

The thesis illustrates the development of DSP-based systems-¡§Hand Written Number Recognition System,¡¨ and ¡§Pattern Recognition System.¡¨ Hand written number recognition system consists of three sub-systems and recognition algorithm: Image Acquisition System, Image Preprocessing System, Image Segmentation System and Binary Pattern Match Algorithm. Pattern recognition system, as well, consists of three sub-systems and recognition algorithm: Image Acquisition System, Image Preprocessing System, Image Segmentation System, and Visual Dynamic Time Warping Algorithm. From the result of the experiment, both DSP image recognition systems can meet the expectation and gain good recognition and efficiency.

No description available.

This work deals with network devices designed for packet routing in small networks. It's focused on two particular devices with ARM processor. There is detailed description of an instalation of OpenWRT operating system to this devices. Next, there are results of tests of routing performance of these and some other selected devices and tests of performance of alogorithms used in network security. A module for OpenWRT operating system for pattern-matching in network traffic is also part of the work.