Grzywna Zbigniew J, Borys Przemysław, Dudek Gabriela
Section of Physics and Applied Mathematics, Faculty of Chemistry, Silesian University of Technology, Ks. M. Strzody 9, 44-100 , Gliwice, Poland.
Biosci Rep. 2006 Apr;26(2):113-29. doi: 10.1007/s10540-006-9011-2.
A set of 10, chosen medicinal plants (some of them with a reputation as remedies for tuberculosis) has been investigated through Partitioned Iterated Function Systems-Semi Fractals with Angle (PIFS-SFA) coding, Lempel, Ziv, Welch with quantization and noise (LZW-QN) compression, and surface density statistics (f(alpha)-SDS) discrimination techniques. The final outcomes of this quantitative analysis were, firstly: the linear ordering of the plants in question accompanied by the hope that it reflects their medical significance, secondly: the mathematical representation of each of the plants, and thirdly: the impressive compression achieved, leading to remarkable computer memory saving, and still permitting successful pattern recognition i.e., proper identification of the plant from the compressed image.
通过分区迭代函数系统-带角度的半分形(PIFS-SFA)编码、带量化和噪声的莱姆佩尔-齐夫-韦尔奇(LZW-QN)压缩以及表面密度统计(f(α)-SDS)判别技术,对一组10种精选的药用植物(其中一些素有治疗结核病的声誉)进行了研究。该定量分析的最终结果如下:其一,相关植物的线性排序,同时希望这能反映它们的医学意义;其二,每种植物的数学表示;其三,实现了令人印象深刻的压缩,显著节省了计算机内存,并且仍能成功进行模式识别,即从压缩图像中正确识别植物。