基于分段拼接的卷烟侧面生成及包灰测定

Generation of cigarette side image and determination of ash integration based on segmented splicing

【目的】建立准确、完整分析卷烟静燃模式下包灰率的方法。【方法】将烟支放置在高精度旋转平台,采集烟灰柱的视频并转换为图像,通过图像预处理、分段拼接、伽马校正和直方图均衡化等处理生成烟灰柱的完整侧面展开图;通过K-means聚类算法进行图像分割并提取裂口区域,统计包灰的像素数和侧面展开图的像素总数,计算两者的比值得到卷烟的包灰率。【结果】利用该方法能够生成单支卷烟燃烧后形成的烟灰柱的完整侧面图,侧面图中的裂口走势连续平滑,实现了对静燃模式下的包灰率的精准评估,所提出的方法优于常规的测定方法。【结论】基于分段拼接而生成的卷烟侧面图像更全面的反映了烟灰柱的裂口情况,对包灰率的测定精度更高,为分析静燃模式下卷烟包灰率提供了一种高精度的测定方法。

[Objective]The ash integration rate is one of the main indexes for evaluating the ash packing performance of cigarettes.The objective of this paper is to accurately and completely analyze this rate under the static combustion mode.[Methods]The cigarettes were placed on a high-precision rotating platform,the video of the ash column was collected and converted into images,a series of processing methods such as image preprocessing,segmented splicing,gamma correction and histogram equalization were used to generate the side expanded image of the ash column;the cracks were extracted by using image threshold segmentation,and the ash integration rate of cigarettes was accurately calculated.This involved counting the number of pixels representing ash integration and the total number of pixels of the side expanded image,and then calculating the ratio of the two to determine the ash integration rate of the cigarettes.[Results]This method is able to generate a complete side expanded image of the ash column after the combustion of a single cigarette,and the fracture trend in the side expanded image is continuous and smooth,which achieves the accurate assessment of the ash integration rate in static combustion mode.The proposed method outperforms conventional measurement techniques.[Conclusion]The side expanded images of cigarettes generated based on segmented splicing more comprehensively reflect the cracks of the ash column,and had higher measurement accuracy for ash integration rate,which provided a high-precision method for analyzing the ash integration rate under the static combustion mode.