基于图像的2维板件排样方案评价方法研究

对于二维不规则件排样的重叠、出界检测和方案评价，传统方法都是使用图形学的方法计算，计算量大，文章提出一种基于图像和颜色统计的新方法。先基于面积和直方图的方法进行了研究，再提出基于边的方法，然后又提出两种改进方法：动态尺度评价和局部重绘整体评价。对每一种方法都进行了时间性能测试和复杂度分析，结果为：基于面积的方法对于小尺寸图像性能尚可，但大尺寸图像时间耗费大，时间为图像尺寸的平方级；基于边的方法时间性能有很大改善，时间耗费为图像尺寸的线性级别；动态尺度评价能基本克服离散方法中尺寸(或精度)与时间(或速度)的矛盾，时间耗费随尺寸的增长低于线性级；局部重绘整体评价则对于大规模排样的时间有大幅改善。最后以三个实例证明了这些方法的有效性。

Research of Layout Estimating Method of 2 D Parts Packing Based on Image

The traditional methods for the problem of the overlap or outside testing and layout estimating in 2-D irregular nesting are all based graphics. This paper brought forward some new methods based on image and color statistics. Firstly put forward a method based on area and histogram, then put forward a better way based on edge, and continually put forward two improving methods: dynamic interval estimating and partially-redraw-all-estimating. Time performance had been tested and complexity had been analyzed for all these methods. The conclusion is: in small image the performance of the method based area is barely OK, but in large image the consuming time is long, being square level of the image dimension; the method based edge has better performance, especially in large image, which time being linear level of image dimension; dynamic interval estimating can basically overcome the contradiction between dimension (or precision) and time (or velocity) in the discrete methods, which time increasing less than linear level; and partially-redraw-all-estimating can shorten the time substantially when parts are many. At last, three examples verified that these methods are effective.