基于图像处理的目标物体最大内接矩形面积的检测

Detecting Maximum Inscribed Rectangle Area of Target Object Based on Image Processing

目标物体最大内接矩形的检测对生产矩形产品的工厂意义重大。目前的裁定手段都是人工实现,存在主观性强、效率低、精度差等缺点。研究了基于图像处理的遍历法、中心扩散法检测目标物体最大内接矩形;并结合两种方法提出一种新的检测方法:遍历中心扩散法。实验结果表明:遍历法对噪声具有很强的鲁棒性;但对凹面物体十分敏感。中心扩散法适用性较好,但检测的准确性不高;而遍历中心扩散法既有较强的鲁棒性,又拥有较高的准确率,其检测准确率均达到90%以上,实现了目标物体最大内接矩形的高精度检测。

It＇s great significant for rectangular products factories to detect the maximum inscribed rectangular of target objects. The traditional method is artificial, with some shortages such as strong subjectivity, low efficiency and poor accuracy etc. In this paper, the traversing and center diffusion methods were utilized to detect the maxi- mum inscribed rectangle of target objects-based on image processing, then on the basis, a new method was pro- posed. The experimental results showed that the traversing was strong robustness to noise, but was sensitive to the concave objects, the applicability of center diffusion was better, but the detection accuracy was lower; while the traversal center diffusion method revealed stronger robustness, and had more than 90% accuracy. Thus the pro- posed detection method in the work is effective for improving maximum inscribed rectangle of the object.