arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 0911.4385cs.CVcs.NE
收藏

仿生速度检测与识别

Bio-inspired speed detection and discrimination

Loria(洛林计算机科学及其应用实验室);INRIA Nancy Grand Est(法国国家信息与自动化研究所南锡大东部分部);Universidad Nacional de Rosario(罗萨里奥国立大学);CIFASIS(阿根廷智能系统研究中心);CONICET(阿根廷国家科学与技术研究理事会)

Mauricio Cerda, Lucas Terissi, Bernard Girau

摘要:

在计算机视觉领域中,一个至关重要的任务是运动检测(也称为光流提取)。该操作允许诸如3D重建、特征跟踪、碰撞时间和新颖性检测等分析。大多数光流提取技术在有限的速度范围内工作。通常,通过在串行架构中组合一些多尺度信息来将检测范围向更高的速度扩展。这种串行多尺度方法受到与算法中使用的尺度数相关的误差传播问题的困扰。另一方面,生物实验表明,人类的运动感知似乎遵循一个并行的多尺度方案。在这项工作中,我们提出了一个生物启发的并行架构来执行运动检测,提供了广泛的操作,并避免与串行架构相关的错误传播。为了测试我们的算法,我们执行经典和建议的技术之间的相对误差比较,表明并行架构能够实现运动检测与串行方法的结果类似。

英文摘要:

In the field of computer vision, a crucial task is the detection of motion (also called optical flow extraction). This operation allows analysis such as 3D reconstruction, feature tracking, time-to-collision and novelty detection among others. Most of the optical flow extraction techniques work within a finite range of speeds. Usually, the range of detection is extended towards higher speeds by combining some multiscale information in a serial architecture. This serial multi-scale approach suffers from the problem of error propagation related to the number of scales used in the algorithm. On the other hand, biological experiments show that human motion perception seems to follow a parallel multiscale scheme. In this work we present a bio-inspired parallel architecture to perform detection of motion, providing a wide range of operation and avoiding error propagation associated with the serial architecture. To test our algorithm, we perform relative error comparisons between both classical and proposed techniques, showing that the parallel architecture is able to achieve motion detection with results similar to the serial approach.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑