A Texture-Based Analysis and Classification of Fruits Using Digital and Thermal Images

A business in the fruit market is totally dependent on the quality of the fruit, its physical appearance, color, shape and size. The fruit market needs fast and effective methods in identifying the worth of the fruits. The present research work employs outer texture of the fruits captured through a digital and a thermal camera. In this present work, eleven different varieties of fruits have been used as main elements for creating a texture-based image database using digital and thermal image capturing devices. The fruit is kept on the revolving tray which can be rotated from 0° to 360° and the entire system is controlled through an interface of stepper motor and the Arduino system and the images are captured from both the image capturing devices. The videos thus captured have been converted to frames for the final creation of the images database having normal RGB and thermal images for further application of image preprocessing techniques. The two image databases upon conversion to gray scale are subjected to texture-based feature extraction technique for studying the outer texture of the fruits. The eleven texture features extracted are subjected to further classification and analysis process for final demarcation of fruits being classified into infected and noninfected grades.