【摘要】 为了实现以较少的内存资源,快速而准确地进行圆检测,提出一种基于粒子群的快速圆检测方法。方法中,以每个粒子作为一个候选圆,采用优化策略,以粒子运动、融合更新等操作完成圆的检测。优化策略包括提出使用一种图像分块的策略限定圆检测的感兴趣区域,各个图像块的大小由各个块内粒子的生成结果决定,以起到抑制部分噪声的作用。为减少生成无用粒子的概率,提高检测效率,提出融入图像金字塔的思想,并以循证的方法对粒子的有效性加以验证。实验结果表明,本文方法对于检测手绘的、嵌套的、不完整的或包含有大量噪声的圆,均能起到较好的检测效果;能够在较短的时间内排除众多干扰,准确地进行圆的检测,具有检测速度快、内存消耗小和适用范围广等优点。更多还原

【Abstract】 In order to detect circles quickly and accurately with less memory,we present a fast circle detection method based on particle swarm optimization(PSO)in this paper.Under some optimization strategies with each particle as a candidate circle,the approach uses some operations for updating the particles to complete the circle detection.The optimization strategies in this approach use an image blocking to limit the circle detection area of interest,and the size of each image block depende on the result of generating particles.In order to reduce the generation probability of those useless particles and improve the efficiency of detection,we use an image pyramid and the evidence-based method to verify the effectiveness of particles.Experimental results validate that the proposed method is suitable for detecting circles which are regular,hand-drawn,nested,incomplete or containing a lot of noise.The proposed method can exclude much interference in a short time,and accurately detect circles.It exhibits small memory consumption,fast running speed and wide application range.更多还原