Comparison of Egg Fertility Identification based on GLCM Feature Extraction using Backpropagation and K-means Clustering Algorithms

This research presents a comparison study of Backpropagation and K-means clustering algorithms for egg fertility identification. Instead of candling the eggs manually, a smartphone camera is used for capturing an egg image, then we do the pre-processing step by performing image enhancement and gray scaling process. The feature extraction method applied in the pre-processed image is the Gray Level Co-occurrence Matrix (GLCM) with six parameters (Entropy, Angular Second Moment, Contrast, Inverse Different Moment, Correlation, and Variance). The result of GLCM's feature extraction image will be processed using two learning algorithms: Backpropagation and K-means Clustering. For evaluation, we use 100 data samples (each in training and testing). The results show that the Backpropagation algorithm (using 12 hidden layer neurons) provides a 93% accuracy rate, while the K-means clustering algorithm presents a 74% accuracy rate. Since the Backpropagation algorithm gives better results in detecting egg fertility, as a recommendation, egg fertility identification can be performed using this algorithm.