Cooperative Cloud-Edge Feature Extraction Architecture for Mobile Image Retrieval

Algorithm 1

Cooperative cloud-edge process.

(i)	Input: image data set, raw image
(ii)	Output: result of image retrieval
(iii)	Cloud gets the projection Matrix A with a KLT algorithm for the image data set storing on the cloud server
(iv)	Cloud sends the projection Matrix A to the edge server
(v)	Edge server gets the raw image uploading from the user equipment
(vi)	Edge server preprocesses the raw image and gets the pixel Matrix , then extracts features by
(vii)	Edge server sends the features of raw image to the cloud server for feature matching;
(viii)	Cloud gets the result of image retrieval with a KNN algorithm
(ix)	Cloud sends the result to the edge server, and the user equipment gets the image retrieval result sending by the edge server