Identification of biodiesel from used cooking oil based on image color characteristics

Abstract

Biodiesel is a biofuel that will be used against the machine or motor type diesel, in the form of ester methyl fatty acids made from vegetable oils or animal. Biodiesel can overcome the problem of the depletion of petroleum and energy crisis. One of the raw materials to make the biodiesel is used cooking oil. Hardware design consists of the design of the black box measuring 27cm x 17cm x 13cm (Length x Width x Height) with 2 pieces of LED as lighting and those ones powered by a 9 Volt battery so that all samples taken in the same conditions. Then the software design consists of designing a GUI in MATLAB. Data retrieval biodiesel utilizing the camera of android SONY Docomo Xperia Z3 which has been equipped with a rear camera 20MP front camera and 5MP. Process to process the image itself with the transformation of the RGB color to HSV to the image by simply selecting the image Hue and the image of the saturation of the course, for the extraction of features (calculate the value of the mean on the image hue and saturation according to the columns or rows of a matrix). The determination of the class using the method of the closest distance that is Euclidean. The first stage to determine the traits that have standard data for reference. and the second stage testing process. Data to determine the characteristic wear 10 of each sample on each of biodiesel, which consists of 3 types. With a total of 30 samples were used as standard data for reference. A system test is performed with the test data, a total of 18 samples only. The results obtained for 15 samples of the test data successfully detected recognizable and 3 sample test data other not successfully detected. The level of accuracy of the system is the introduction of biodiesel shows the results of 83.3% by using the method of Euclidean distance, which means the level of accuracy is high.