基于图像轮廓处理及匹配的木节剔除和修复技术

Image Contour Processing and Matching-based Knot Removal and Repair Technology

采用计算机图像处理技术实现板材缺陷(木节)的剔除和修复时,首先对含有木节的板材进行图像处理,通过灰度化和滤波去除噪声,再采用二值化阈值法将木节从板材图像中提取出来,对提取出的木节图像进行膨胀、腐蚀、边缘检测等处理,提取出木节的轮廓边界。根据缺陷的横向和纵向跨度确定出修补中心点及修补半径。为了实现补块与剔除区域边缘的图像匹配,提出了基于灰度的环形块匹配算法。从剔除区域取一窄圆环,并将其等分为几个小块,按顺时针对各小块进行编号,计算出各环形小块的平均灰度值,用各小块的平均灰度值代替单个元素进行与数据库中补块的各小块灰度值进行匹配,并选相似度最大者为匹配补块,这一算法能快速有效地找到匹配补块及匹配位置。该方法有助于实现木材缺陷的自动检测和自动修复,对提高板材的利用率具有现实意义。

When computer image processing technology is used to remove and repair wood detects（knots）,the image of the wood sheet containing defects（knots）is processed first,the noise is eliminated through gray processing and filtration and the knot is extracted from the wood image through the binary threshold method,followed by expansion,corrosion,boundary detection treatment of the extracted knot image.According to the horizontal span and vertical span of the defect,the central point for repair and repair radius are determined.In order to realize the edge image matching of the repair block and the area where the knot is removed,a ring block matching algorithm is proposed based on the gray value.A narrow ring is selected from the removed area and divided into several small equal blocks,each of which is numbered clock-wisely,the average gray value of each ring block is calculated.The average gray value of each block will replace a single element to match the gray value of each small block in the database used for repair and the one with most similarity will be chosen as the matching repair block.The algorithm can quickly and effectively find the matching repair block and matching position.The method helps realize the automatic detection and repair wood defects improve the utilization rate of wood.