灰度直方图在木材表面缺陷检测中的应用

Application of grey histogram to wood surface defects' detection

提出了一种基于灰度直方图检测木板材表面缺陷的方法。对图像以4×4像素块作25级灰度直方图分析,根据灰度直方图统计中是否有颜色突变来判别木板材是否存在缺陷。缺陷图片在直方图中表现出双峰特征,利用次波峰确定缺陷部位,但判别时应排除杂色与纹理因素的干扰。比较次波峰与主波峰的值来消除木材杂色所产生的影响,即当差值大于1/10时,次波峰代表缺陷颜色。与此同时,借助于直方图修正排除纹理因素造成的干扰。该方法能有效地识别缺陷图像与正常图像。

A method for the detection of defects in the wood surface by grey histogram, which was analyzed by 25 grades for every 4 × 4 block pixels, was studied in the research. It judged the existence of defects in the woods according to the color break in grey-histogram. Defect pictures had double peaks in the grey-histogram and usually the subordinate peak was the defect part. However, the influential factors including motley and texture were not considered during the first step. The subordinate peak and host peak values were compared to eliminate the disturbance caused by motley. The research findings showed that the subordinate peak value indicated defect color if it was 1/10 times greater than the value of the host peak. Meanwhile, the disturbance caused by texture could also be eliminated through histogram amendment. [ Ch, 8 fig. 10 ref. ]