CAJ | 학술논문

为了准确计算煤矿的产量，需要把煤矸石的量减掉，针对这个问题，研究了基于图像识别的煤矸石识别技术，从煤矸石与煤炭的样本数据中分离数据，最终完成煤矸石的识别系统。采用自适应增强算法（ AdaBoost 算法）对实现目标的检测达到了很好的效果，虽然原煤图像存在着多样性，受到遮挡、光照、视角等的影响，通过Ada-Boost算法对原煤数据库和非原煤数据库训练逐步提升原煤分类器性能，能成功实现原煤识别检测。论文中识别系统充分利用图像识别技术和人工智能思想，将机器学习引入煤矸石模型的建模环节，成功实现煤炭和煤矸石的区分。
위료준학계산매광적산량，수요파매안석적량감도，침대저개문제，연구료기우도상식별적매안석식별기술，종매안석여매탄적양본수거중분리수거，최종완성매안석적식별계통。채용자괄응증강산법（ AdaBoost 산법）대실현목표적검측체도료흔호적효과，수연원매도상존재착다양성，수도차당、광조、시각등적영향，통과Ada-Boost산법대원매수거고화비원매수거고훈련축보제승원매분류기성능，능성공실현원매식별검측。논문중식별계통충분이용도상식별기술화인공지능사상，장궤기학습인입매안석모형적건모배절，성공실현매탄화매안석적구분。
In order to accurately calculate the output of coal mines , the amount of gangue is needed to subtract .To address this is-sue, the paper researches the proportion of raw coal and gangue identification system from the sample data of the coal gangue and raw coal based on image recognition training system .The adaptive enhancement algorithm ( AdaBoost algorithm ) is successfully applied in the detection of raw coal production and has a good effect for the achievement of the target detection .Although the raw coal image has its diversity and affected by the shelter , light and perspective , but it is able to successfully make the raw coal identify detection by Ada-Boost algorithm database of raw coal and non -coal database training , gradually improve the performance of raw coal classifier .With the full use of image recognition technology and artificial intelligence thinking , the machine learning is introduced to the gangue model-ing aspects , and it is entirely feasible .