Toward automatic quality detection of Jasmenum flower

In this paper, a novel algorithm is proposed for the classification of flowers, such as petal deformation intra and inter class variability, and illumination and detection of their quality. Several approaches could be incorporated to extract the flower features; however the proposed algorithm highly deals with the quality detection in various aspects specific to the Jasmine flower. The proposed algorithm focuses on the color, shape and texture features, combined to identify the flower quality. In the first phase of implementation, the color features were extracted using the average color difference and color and edge directivity descriptor method. In the second phase of the implementation, the shape and texture features were extracted using Zernike moments and local binary patterns. The third phase involved the classification by using a support vector machine and random forest tree classifiers. The performance of the algorithm was verified by comparing it with the trained and existing data.