Performance Analysis of Cooperative Hopfield Networks for Stereo Matching

Authors:
Wenhui Zhou;Zhiyu Xiang;Weikang Gu
Affiliations:
Dept.Information Science and Electronic Engineering, ZheJiang University, HangZhou, 310027, Email: sunshine@hzcnc.com, China;Dept.Information Science and Electronic Engineering, ZheJiang University, HangZhou, 310027, Email: sunshine@hzcnc.com, China;Dept.Information Science and Electronic Engineering, ZheJiang University, HangZhou, 310027, Email: sunshine@hzcnc.com, China
Venue:
Computational Intelligence and Security
Year:
2007

Citing 6
Cited 0

Edge Detection with Embedded Confidence

IEEE Transactions on Pattern Analysis and Machine Intelligence
Mean Shift Analysis and Applications

ICCV '99 Proceedings of the International Conference on Computer Vision-Volume 2 - Volume 2
What Energy Functions Can Be Minimizedvia Graph Cuts?

IEEE Transactions on Pattern Analysis and Machine Intelligence
An Experimental Comparison of Min-Cut/Max-Flow Algorithms for Energy Minimization in Vision

IEEE Transactions on Pattern Analysis and Machine Intelligence
Neural adaptive stereo matching

Pattern Recognition Letters
Multilevel- and neural-network-based stereo-matching method for real-time obstacle detection using linear cameras

IEEE Transactions on Intelligent Transportation Systems

Quantified Score

Hi-index	0.00

Visualization

Abstract

This paper proposes a dense stereo matching algorithm based on cooperative Hopfield networks. It uses two Hopfield networks with similar structure to solve energy minimization problem of stereo matching in parallel. Two strategies are applied to the performance analysis. One strategy considers each pixel as a neuron. The other is the Coarse-to-Fine strategy, which firstly divides the images into non-overlapping homogeneous regions, and each region is represented as super-pixel of the coarse images. After coarse estimation, a more refined estimation is implemented in pixel domain. Experiments indicate the method with the Coarse-to-Fine strategy has better performance and more rapid convergence speed, and less insensitive to initial conditions of the neural networks and the neuron update orders.