应用于刹车盘复杂背景去除的改进Grab Cut分割算法

Improved Grab Cut Segmentation Algorithm for Removing the Complex Background of Brake Disc

针对无序分拣过程中车间光照不均导致的分割结果不准确问题,提出一种改进的Grab Cut图像分割算法。该算法采用自适应中值滤波去除图像表面噪声;通过限定对比度自适应直方图均衡化提升图像整体亮度,增强对比度;进一步使用Laplacian算子对刹车盘区域进行锐化处理;在此基础上对刹车盘图像复杂背景进行去除;最后通过仿真实验对比分割效果、PSNR、误分割率、分割准确性和运行时间来评价算法模型。实验结果表明:提出的改进Grab Cut分割算法相较于原始算法能有效降低光照不均对刹车盘复杂图像背景去除的影响,使分割准确率提高了12.5%,分割结果与实际刹车盘模型相比误差小,为刹车盘图像复杂背景去除提供参考。

An improved Grab Cut image segmentation algorithm was proposed to solve the problem of inaccurate segmentation re⁃sults caused by uneven lighting in the workshop during unordered sorting.In the algorithm,adaptive median filter was used to remove the image surface noise.By limiting contrast adaptive histogram equalization,the overall brightness of the image and the contrast were enhanced;Laplacian operator was further used to sharpen the brake disc area;on this basis,the complex background of the brake disc image was removed.Finally,the algorithm model was evaluated by comparing the segmentation effect,PSNR,segmentation er⁃ror rate,segmentation accuracy and running time through simulation experiments.The experimental results show that compared with the original algorithm,the proposed improved Grab Cut segmentation algorithm can be used to effectively reduce the influence of une⁃ven light on the removal of complex image background of the brake disc,and the segmentation accuracy is increased by 12.5%.The segmentation result has a small error compared with the actual brake disc model,providing reference for the removal of complex image background of the brake disc.