Kombinasi Metode Discrete Cosine Transform Dan Convolutional Neural Network Dalam Mengidentifikasi Tingkat Kematangan Buah Mangga Berdasarkan Warna

Abstract

The classification of mango fruit ripeness levels is currently predominantly done manually, which unfortunately has several drawbacks. One of the primary shortcomings is the lack of consistency in accuracy, often resulting in differences among operators conducting the sorting. On the other hand, in image classification processes, a combination of the Discrete Cosine Transform (DCT) and Convolutional Neural Network (CNN) methods is utilized. DCT is a technique commonly used in image processing, especially for image pictures. In this research, there is a proposal to merge the Discrete Wavelet Transform method with the Convolutional Neural Network (CNN). Currently, CNN is one of the methods that provides the most significant results in image recognition. CNN attempts to mimic the image recognition system in the human brain, particularly the visual cortex, allowing it to efficiently process image information. The DCT method is used to transform image data into a frequency image form, which is subsequently employed in feature extraction in the Deep Neural Networks classification method. The research results indicate that the combined method of Discrete Cosine Transform and Convolutional Neural Network achieves the highest accuracy rate of 93.33% in classifying mango fruit ripeness levels. This outcome demonstrates significant potential for automating the mango ripeness classification process with high accuracy, overcoming the inconsistencies associated with manual approaches.