基于图割算法的木材表面缺陷图像分割

Wood surface defect image segmentation based on Graph Cut algorithm

木材表面缺陷分割的研究能够有效提高木材的利用率,节约现有木材资源,缓解森林资源短缺的压力。为了更好地对板材表面的节子和虫眼进行快速有效地分割,论述了基于图割算法的图像分割方法(Graph Cuts)及其改进方法(Grab Cuts)的原理。针对传统Graph Cuts算法只能针对灰度图像进行分割、运行时参数的选择比较复杂,并且存在该算法效率和精度较低的缺陷,采用这两种方法分别对3种木材表面缺陷活节、虫眼和死节图像进行分割实验。为了验证Grab Cuts方法的适用性,用含有多个缺陷目标的木质板材图像做了图像分割验证。结果表明:缺陷图像的目标和背景的种子点选取直接影响Graph Cuts算法的分割结果,Graph Cuts算法的计算效率较低,分割时间较长,对相邻像素间的区分度较差,分割结果不理想。改进后的Grab Cut算法是迭代的Graph Cuts,该方法虽然在图像分割前也需要人工画定初始化矩形框,但操作相对简单,分割结果能够得到完整的闭合缺陷区域边界,且不受木材表面缺陷的类型、数量、尺寸和缺陷形状的影响,分割效果好,分割速度快,抗噪性强,对灰度图像和彩色图像都可使用。

Since wood surface defect can seriously affect its performance,quality and use value,the segmentation and detection of wood surface defects play important roles in lumber grading evaluation and quality control.Slipknot,wormhole and encased knot of wood surface defect images are selected as the object of this study.In this paper,we investigated the single object,the wood surface defect of multipletarget gray image and color image segmentation.By analyzing the data,we know that wood surface defect image segmentation can effectively improve the utilization rate of wood,thus saving precious forest resources.In order to segment knot and wormhole on plankssurfaces rapidly and effectively,we discussed the principle of the Graph Cut method,which is only effective for gray image segmentation,complex in choosing runtime parameters,and low efficiency and accuracy,and the improved Grab Cut method.These two methods are used to conduct a segmentation experiment with regard to the images of three wood surface defects,namely,slipknots,wormholes and knots.In order to explore the applicability of the Grab Cuts method,the image segmentation verification was performed by using a wood plate image with multiple defect targets.The results showed that the selection of seed points in the target and background of the defect image directly affects the segmentation result of Graph Cuts algorithm.The segmentation result is not ideal because of the low calculation efficiency of Graph Cuts algorithm,long segmentation time,and poor discrimination between adjacent pixels.The improved Grab Cut algorithm is an iterative Graph Cuts.Although this method requires manual drawing of the initialized rectangular frame before the image segmentation,the operation is relatively simple,and the segmentation result can achieve a completely closed boundary of the defect area and is not affected by the type,quantity and size of surface defect,and the shape of defects.Due to the advantages of good segmentation effect,fast segmentation and strong antinoise,this method is suitable for both grayscale and color images.