A MODULAR AND SCALABLE VLSI DESIGN FOR ROBUST OBJECT DETECTION AND TRACKING IN UNPREDICTABLE ENVIRONMENTS

Abstract

Abstract – The development of digital image processing has been a significant contributor to the improvement of machine perception and the interpretation of complex scenes, particularly in environments that are dynamic and unpredictable. Applications such as autonomous navigation, surveillance, remote sensing, medical imaging, and
industrial automation are just some of the many that require robust, real-time object detection and tracking systems. These kinds of systems are necessary in light of the circumstances. An adaptive Very Large
Scale Integration (VLSI) architecture that is specifically tailored for robust object detection and tracking in dynamic environments is the goal of this study, which aims to propose such an architecture. To effectively manage variable resolution and rapid scene changes while maintaining computational efficiency, the proposed system incorporates a space-variant edge detection mechanism and an adaptive
image scaling unit. This makes it possible for the system to effectively deal with both of these kinds of factors. Interpolation-induced blurring can be reduced, and edge precision can be improved through the use of
a pre-filtering technique that makes use of spatial filters. Additionally, improved detection accuracy is achieved through the implementation of hardware-efficient bilinear interpolation in conjunction with adaptive pixel selection.