東華大學圖書館 |

語系: 繁體中文

說明(常見問題)

回圖書館首頁

手機版館藏查詢

登入

回首頁

切換: 標籤 | MARC模式 | ISBD

Noninvasive classification of embryo...

Chalker, Brad Alan, II.

FindBook

Google Book

Amazon

博客來

Noninvasive classification of embryonated poultry eggs using visible and near-infrared spectral analysis.

紀錄類型:	書目-電子資源 : Monograph/item
正題名/作者:	Noninvasive classification of embryonated poultry eggs using visible and near-infrared spectral analysis./
作者:	Chalker, Brad Alan, II.
面頁冊數:	119 p.
附註:	Source: Dissertation Abstracts International, Volume: 66-03, Section: B, page: 1568.
Contained By:	Dissertation Abstracts International66-03B.
標題:	Engineering, Biomedical. -
電子資源:	http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=3170412
ISBN:	0542065207

Noninvasive classification of embryonated poultry eggs using visible and near-infrared spectral analysis.
Chalker, Brad Alan, II.

Noninvasive classification of embryonated poultry eggs using visible and near-infrared spectral analysis. - 119 p.

Source: Dissertation Abstracts International, Volume: 66-03, Section: B, page: 1568.

Thesis (Ph.D.)--The University of North Carolina at Chapel Hill, 2005.

Candling has developed into a common process in the poultry industry by which the internal qualities of an egg may be noninvasively determined. When held up to a bright light, the contents of an egg are illuminated and can be visually characterized. Automated candling devices exist, but are in limited use in hatching egg facilities due to limitations in their classification capabilities. In other fields, more sophisticated devices analyze the intensity of light at discrete wavelengths across a spectrum. The hypothesis of this dissertation was that spectral analysis could be used to noninvasively classify embryonated eggs. Egg classes analyzed included: alive, infertile, early dead, middle dead, rotten, inverted alive, and cracked-dried out. Initial project steps included determining a preferred apparatus configuration and collecting a large set of visible and near-infrared spectra. With a training subset, optimal classification wavelengths were calculated using correlation coefficients and linear regression, and then used to generate two dimensional dot plots. Cutoff linear formulas to separate the classes were calculated from the dot plots and then combined into a scoring system that yielded class scores for an unknown spectrum. A validation subset of spectra was then run through the scoring system, which resulted in accuracies ranging from 80.4% for early dead eggs to 97.3% for rotten eggs. Next, in order to mimic automated candling systems, which would use discrete photodiodes and light sources, a narrowed selection of eight different wavelengths was chosen, based upon the distribution of the wavelengths in the original analysis. The algorithm development process was repeated with only these wavelengths, which resulted in accuracies ranging from 70.1% for early dead eggs to 93.4% for inverted alive eggs. Finally, experiments were conducted to look at the effects of the shell on the spectra collected and classification accuracies. Possible biological and physical causes for the clustering of key analysis wavelengths in the range 650 nm to 825 nm were examined, along with potential future work to improve the hardware configuration or developed algorithms. The conclusion was that spectral analysis was capable of classifying embryonated eggs with statistically significant accuracy.

ISBN: 0542065207Subjects--Topical Terms:

1017684
Engineering, Biomedical.

Noninvasive classification of embryonated poultry eggs using visible and near-infrared spectral analysis.
LDR:03268nmm 2200277 4500 001 1849893
005 20051203081304.5
008 130614s2005 eng d
020 $a 0542065207
035 $a (UnM)AAI3170412
035 $a AAI3170412
040 $a UnM $c UnM
100 1 $a Chalker, Brad Alan, II. $3 1937827
245 1 0 $a Noninvasive classification of embryonated poultry eggs using visible and near-infrared spectral analysis.
300 $a 119 p.
500 $a Source: Dissertation Abstracts International, Volume: 66-03, Section: B, page: 1568.
500 $a Adviser: Richard L. Goldberg.
502 $a Thesis (Ph.D.)--The University of North Carolina at Chapel Hill, 2005.
520 $a Candling has developed into a common process in the poultry industry by which the internal qualities of an egg may be noninvasively determined. When held up to a bright light, the contents of an egg are illuminated and can be visually characterized. Automated candling devices exist, but are in limited use in hatching egg facilities due to limitations in their classification capabilities. In other fields, more sophisticated devices analyze the intensity of light at discrete wavelengths across a spectrum. The hypothesis of this dissertation was that spectral analysis could be used to noninvasively classify embryonated eggs. Egg classes analyzed included: alive, infertile, early dead, middle dead, rotten, inverted alive, and cracked-dried out. Initial project steps included determining a preferred apparatus configuration and collecting a large set of visible and near-infrared spectra. With a training subset, optimal classification wavelengths were calculated using correlation coefficients and linear regression, and then used to generate two dimensional dot plots. Cutoff linear formulas to separate the classes were calculated from the dot plots and then combined into a scoring system that yielded class scores for an unknown spectrum. A validation subset of spectra was then run through the scoring system, which resulted in accuracies ranging from 80.4% for early dead eggs to 97.3% for rotten eggs. Next, in order to mimic automated candling systems, which would use discrete photodiodes and light sources, a narrowed selection of eight different wavelengths was chosen, based upon the distribution of the wavelengths in the original analysis. The algorithm development process was repeated with only these wavelengths, which resulted in accuracies ranging from 70.1% for early dead eggs to 93.4% for inverted alive eggs. Finally, experiments were conducted to look at the effects of the shell on the spectra collected and classification accuracies. Possible biological and physical causes for the clustering of key analysis wavelengths in the range 650 nm to 825 nm were examined, along with potential future work to improve the hardware configuration or developed algorithms. The conclusion was that spectral analysis was capable of classifying embryonated eggs with statistically significant accuracy.
590 $a School code: 0153.
650 4 $a Engineering, Biomedical. $3 1017684
650 4 $a Agriculture, Food Science and Technology. $3 1017813
690 $a 0541
690 $a 0359
710 2 0 $a The University of North Carolina at Chapel Hill. $3 1017449
773 0 $t Dissertation Abstracts International $g 66-03B.
790 1 0 $a Goldberg, Richard L., $e advisor
790 $a 0153
791 $a Ph.D.
792 $a 2005
856 4 0 $u http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=3170412