MODEL FOR IDENTIFICATION OF RICE TYPE USING COMBINATION OF SHAPE AND COLOR FEATURES Jumi (a*), Achmad Zaenuddin (b), Tedjo Mulyono (b)
a)Bussiness Administration Departement, Politeknik Negeri Semarang, Central Java, Indonesia b)Bussiness Administration Departement, Politeknik Negeri Semarang, Central Java, Indonesia b)Civil Engineering Departement, Politeknik Negeri Semarang, Central Java, Indonesia
The development of applied knowledge in the field of technology that advances industry in society is the master plan of Higher Education research, one of the areas of focus is the development of information technology. Information technology is needed in almost all fields of industry, including food industry, namely rice. That rice is one of the leading agricultural commodities in Indonesia which has various levels of quality and type. The many types and levels of quality of rice require a database and precise and consistent identification in classifying it. Determination of the level of quality and type of rice can be done using visual data, namely rice imagery. Each type of rice has relatively different physical characteristics, as well as each level of quality of rice has different physical characteristics. Through physical characteristics in the image that is shape, color and texture will be found in the level of quality and type of rice. Characteristics or physical features of rice stored in the image are strongly influenced by conditions when taking rice images such as lighting, the position of the camera and the distance of the camera to the object which will further affect the results of identification. This requires image processing methods to improve detection accuracy so that the identification of quality and type of rice has a high level of validation. This research has developed a model of identification of rice types using several image features including shape features, color features and a combination of the two features. The results of testing using this model are proven to produce more than 60% accuracy for the use of shape features only, while identification using only the color features has an accuracy of more than 55%. Identification using a combination of shape and color features has an accuracy of more than 75%.