This preview has intentionally blurred sections. Sign up to view the full version.View Full Document
Unformatted text preview: Enhanced Image Detection on an ARM based Embedded System J.R.EVANS [email protected] University of Edinburgh, Department of Electronics and Electrical Engineering, King's Buildings, Mayfield Road, Edinburgh EH93JL, Scotland, UK T. ARSLAN University of Edinburgh, Department of Electronics and Electrical Engineering, King's Buildings, Mayfield Road, Edinburgh EH93JL, Scotland, UK Abstract. This paper presents a new technique for the detection of Integrated Circuits within images of Printed Circuit Boards autonomously and without the need to be assisted by CAD data. The technique is a key part of a suite of algorithms targeted for an embedded System On Chip architecture based on the ARM7 platform for real time detection of PCB images for diagnostic purposes. The technique has a significant reduction in complexity when compared to conventional approaches such as the Hough Transform. The reduction in complexity makes the approach ideal for an embedded vision application such as the one described in this paper. This paper presents the technique, the target embedded architecture and results showing the reduction in complexity when compared to a Hough Transform. Keywords: Automatic optical inspection, image processing, system on chip. 1. Introduction With the advent of System On Chip (SOC) technology it is now possible to integrate complex software and/or hardware functionality on a single chip.Currently such systems are being targeted for real time applications where real time processing is a crucial issue. For such applicationsthe software components are usually the most performance critical. For this reason it is important that such components are highly optimised in terms of speed while using a relatively small section of the available embedded memory. Imagerecognition isan importantpartoftheMachineVision field.Objectdetection isa important class of problem within Image recognition. Object recognition and manipulation algorithms are characterised by being computationally complex due to the size of both image and source system added to the large number of complex arithmetic operations. It is extremely desirable that such applications are performed on a standard SOC embedded processor without the need for large and expensive memories and co- processors. The main aim of this paper is therefore to show the feasibility of implementing object recognition algorithms on a standard SOC target which provide the advantage of flexibility in addition to that of real time speed. Effective methods of object detection of Integrated Circuits (ICs) are of considerable interest to developers of Automatic Optical Inspection (AOI) systems for analyzing Printed Circuit Boards (PCBs). AOI systems are becoming more important in the Design Automation for Embedded Systems,6,477^487, 2002....
View Full Document
This note was uploaded on 11/03/2009 for the course COMPUTERS CS537 taught by Professor Salman during the Spring '09 term at Texas A&M University–Commerce.
- Spring '09