Identifikasi embrio dalam telur berbasis image processing

Abstract

One of ways to increase the success of hatching eggs is by selecting and separating the eggs embryonated (fertile) with eggs are not embryonated (infertile) by way of observation (candling). This system utilizes digital image processing as an identification process. By this system, it is expected that the identification results will be more accurate results than conventional monitoring, so as to increase the results of hatching. This system utilizes a flashlight as a medium, so that the egg's internal condition can be seen which then takes pictures by the webcam. After that the digital image processing is done by converting the original image (RGB) to binary image by providing a thresholding value (T), the T value is very influential in the next image processing, opening and closing, thinning the image (thinning), and contour detection. Then from the final process of contour detection produces the number of detection of blood vessels that are considered as embryos as a determinant of the outcome category of identification. From the experiments carried out the percentage of conformity between the original condition of the egg with the results obtained in the system that is 88.88%, in determining the yield category (fertile/infertile) with an error of 11.12%. For the suitability of the estimated percentage of hatchlings themselves have a success of 61.11% with an error of 38.89%. These results are influenced by many factors like the condition of the eggs and supporting devices in the system.