Abstract
Hough transform (HT) is one of the most well-known techniques in computer vision that has been the basis of many practical image processing algorithms. HT however is designed to work for frame-based systems such as conventional digital cameras. Recently, event-based systems such as Dynamic Vision Sensor (DVS) cameras, has become popular among researchers. Event-based cameras have a significantly high temporal resolution (1 μs), but each pixel can only detect change and not color. As such, the conventional image processing algorithms cannot be readily applied to event-based output streams. Therefore, it is necessary to adapt the conventional image processing algorithms for event-based cameras. This paper provides a systematic explanation, starting from extending conventional HT to 3D HT, adaptation to event-based systems, and the implementation of the 3D HT using Spiking Neural Networks (SNNs). Using SNN enables the proposed solution to be easily realized on hardware using FPGA, without requiring CPU or additional memory. In addition, we also discuss techniques for optimal SNN-based implementation using efficient number of neurons for the required accuracy and resolution along each dimension, without increasing the overall computational complexity. We hope that this will help to reduce the gap between event-based and frame-based systems.
Highlights
Neuromorphic engineering is an inter-disciplinary field focusing on implementing the biological neural systems on software and hardware systems such as analog, digital, or mixed-mode electronic circuits
In this paper we systematically discussed efficient implementation of Hough Transform (HT) for event-based systems, the concepts that should be considered for each application, and the challenges
As the Standard HT (SHT) cannot be readily applied to event-based systems, as it is structured for frame-based input streams
Summary
Neuromorphic engineering is an inter-disciplinary field focusing on implementing the biological neural systems on software and hardware systems such as analog, digital, or mixed-mode electronic circuits. Bachiller-Burgos et al (2018) presents a 3D Hough transform model for detecting corner points using a 3D SNN. We focus our discussion on implementation of Hough Transform via a two- or three-dimensional SNN to find all locally linear elements in a video captured in an event-based input stream (e.g., from a DVS camera). These elements can be utilized to obtain shape information of a free-form curve, and to run other possible post-processing steps for shape encoding, extraction, etc. We discuss in detail uniform and non-uniform parameter space quantization and their relation to different application settings followed by conclusion and future works
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have