Evolutionary Computing and Particle Filtering: A Hardware-Based Motion Estimation System

Abstract

Abstract-Particle filters constitute themselves a highly powerful estimation tool, especially when dealing with non-linear non-Gaussian systems. However, traditional approaches present several limitations, which reduce significantly their performance. Evolutionary algorithms, and more specifically their optimization capabilities, may be used in order to overcome particle-filtering weaknesses. In this paper, a novel FPGA-based particle filter that takes advantage of evolutionary computation in order to estimate motion patterns is presented. The evolutionary algorithm, which has been included inside the resampling stage, mitigates the known sample impoverishment phenomenon, very common in particle-filtering systems. In addition, a hybrid mutation technique using two different mutation operators, each of them with a specific purpose, is proposed in order to enhance estimation results and make a more robust system. Moreover, implementing the proposed Evolutionary Particle Filter as a hardware accelerator has led to faster processing times than different software implementations of the same algorithm.

High-Performance Parallel Implementation of Genetic Algorithm on FPGA

“…As well as the high-performance achieved, the small area consumption of the implementation developed here is a notorious feature. This makes it possible for other systems to also be embedded in the FPGA, since the on-board GA occupies less than 1 5 of the Virtex 7 logic cells used as a test. This logical cells low consumption feature is essential for applications where the area is the biggest constraint as spatial applications, for example.…”

“…One way found by researchers and developers to address such demands is by using algorithm parallelization techniques. Parallel processing is used to manipulate data concurrently, so that while computing one section of the algorithm, other stations perform similar operations on another set of data [1]. Combining the hardware implementation with the parallelization of algorithms is often a satisfactory solution for high performance and higher speed applications when compared to sequential solutions.…”

VLSI Implementation of Motion Estimation Using Level Convertor with CSLA Adder

“…Most of the existing algorithms used for ME are Bilateral ME [10], ego ME [11], variable block size ME [12], optical flow estimation [13], full search ME [14], and particle ME [15]. But, these kind of methods doesn't have any motion pattern that affects the quality of the video.…”