SCOTECT ALGORITHM: A NOVEL APPROACH FOR SOIL COLOR DETECTION PROCESS USING FIVE STEPS ALGORITHM

摘要

Soil is a major component of the land. Soil color is often used as the initial impression when we view the soil. The color is also affected by the environment, deepness, mineral content, etc. Munsell Soil Color Chart is a book to classify soil colors, but so difficult to classify the color of soil using this book. In our paper, we research how soil color can be classified using the proposed algorithm, Soil Color Detection (Scotect) algorithm. Scotect is five steps algorithm to detection the color of the soil. The main process is to detect the color of the soil. The first step is creating the database, and we can use mode of RGB value to get representation data. Second, a median filter method is used to get the clearer image. Third, an image will be segmented by using K-means segmentation method. Furthermore, the segmented image will be filtered again by using median filter method. And the last process is matching each layer of image soil with color in the database using Euclidean distance. This research succeeded in finding the new way for detecting the color of the soil. We succeed in showing that the program can segment soil image. The most important is that this algorithm's output succeeded concluding that result of this program is 90.58% accurate to retrieve label of the testing data.