A Color Constancy Based Flower Classification Method in the Blockchain Data Lake

The efficient classification of flower images will directly affect the accuracy of their automatic recognition. Due to the complexity of the background of flowers, not only the color, shape and texture of flowers are different, but also the illumination factors show significant effect on classification results of flower images during the process of acquiring flower images. Therefore, it is of great practical significance to identify flowers with the help of flower salient features and eliminate lighting factors. In order to reduce the influence of illumination factor on the classification accuracy of flower images and ensure the true transparency of flower images in the process of Internet data transmission, in this paper, we propose a color constancy based flower classification method in the Blockchain Data Lake, short for CCAN, firstly, we design a Blockchain Data Lake framework to ensure the accuracy and originality of the original image data; and then, color constancy mechanism is used to encode the color feature of images, in order to reduce the illumination effects. Thirdly, a convolutional neural network based classifier is proposed to achieve flower classification. Finally, we simulate the performance of CCAN on three different data set in the blockchain Data Lake environment, extensive results show that the proposed CCAN effectively improves the accuracy of flower image classification by minimizing the interference of illumination factors on flower targets.