摘要
为了在蛋品加工中能够快速地检测并剔除破损蛋,将声卡采集到的蛋在敲击时的声脉冲信号送入计算机,利用倒谱和功率谱进行分析,分别提取出6个特征参数,并进行逐步判别分析,选择出最优判别因子,建立判别模型。实践表明:其对破损蛋的识别准确率达到95%,对禽蛋破损检测提供了参考方法。
In order to quickly find out and pick out cracked-shell eggs in the processing of eggs, the acoustics signals created by striking the eggshells are acquired by a sound card and transformed into digital signals. Through analyzing those signals, six characteristic parameters were extracted respectively from power spectrum and cepstrum. Step discriminance was used to optimize the combination of these characteristic parameters. Establish the optimum distinguishing functions of normal eggshells and cracked eggshells respectively. The accurate recognition ration for cracked-shell eggs is up to 95.5 percent.
出处
《农机化研究》
北大核心
2007年第11期192-194,共3页
Journal of Agricultural Mechanization Research
基金
湖南工程学院硕博基金项目(120641)
关键词
食品丁业
禽蛋破损检测
应用
声音信号
倒谱
功率谱
food industry
egg crack detection
application
acoustic signal
cepstrum
power pectrum
