摘要
通孔和盲孔图像差异大,可根据图像计算出各灰度值的概率分布,通过灰度直方图完成孔型识别和二值化阈值匹配。由像素点概率分布计算各灰度值概率,作出反映灰度值像素点在图像中所占比例的孔型像素灰度直方图。突出通孔和盲孔灰度值集中部分的百分比,得到孔的直方图。二值化阈值匹配,通孔按灰度直方图算出的两峰间最佳阈值得到,而盲孔用大津法求解。
The difference is big between the through-hole and blind hole, so the probability of the gradation value is calculated in accordance with the image, and the pass recognition and automatic matching of threshold value is finished through the gradation histogram. The probability of the graduation value is calculated by the probability of the pixels, and the pass pixels gradation histogram reflected the proportion of the graduation value pixels in the image is drawn. The through-hole histogram is gained by standing out the concentrating percent of gradation value of the through-hole and blind hole. About automatic matching of threshold value, the prime threshold value of the through-hole is gained by the histogram, and the prime threshold value of the blind hole is calculated by Otsu.
出处
《兵工自动化》
2005年第2期48-49,52,共3页
Ordnance Industry Automation
基金
四川省科技厅基金项目04GY029-003-6
关键词
图像识别
孔型识别
灰度直方图
阈值
Image recognition
Pass recognition
Gradation histogram
Threshold value
