摘要： 为实现枪弹外观颜色缺陷快速、自动、准确分割，提出一种基于混合颜色模型的枪弹外观缺陷分割方法。分析统计正常光照条件下弹体和颜色缺陷图像特征，获得统计规律，提取颜色特征，分割颜色缺陷，对感兴趣的局部弹体图像进行由RGB到HSI的颜色模型转换，利用色调H分量进一步分割缺陷，使用数学形态学优化分割结果，定义缺陷平均分割率和平均偏差量化分割结果。实验结果表明，该方法在弹体缺陷分割的应用中表现出较好的准确性和鲁棒性。

关键词: 外观缺陷, 图像分割, 颜色模型, 阈值, 数学形态学

Abstract: In order to segment the bullet appearance color detect quickly, automatically and accurately, a new method of image segmentation is proposed based on hybrid color model. The color feature detect is segmented, which is extracted and analyzed from image feature. Aiming at the shortcoming of image segmentation by RGB color model under the condition of strong illumination, convert the color model from RGB to HSI in the interesting region of image, the detect is segmented by H component further. The result of segmentation is optimized by mathematical morphology and evaluated by average segmentation rate and average deflection. Experimental result shows that this method has extrusive accuracy and robustness in the application of bullet surface defect segmentation.

Key words: appearance defect, image segmentation, color model, threshold, mathematical morphology

中图分类号: 

• TP391.4