Sketch Based Image Retrieval Based on Chamfer Distance Transform and Bag of Mid Maps Descriptor

With the popularity of touch-screen devices, sketch based image retrieval has attracted more and more attention. Considering the limitation of the traditional descriptors such as HOG, SIFT and RST-HELO, we proposed a novel feature descriptors based on the bag of mid maps of the convolutional neural network. Our work is realized by the following steps:1) Extracting the boundary probability images of the database; 2) Converting the boundary probability images into Chamfer distance images; 3) Generating the bag of mid maps descriptor for final retrieval. We evaluated our proposed descriptor and retrieval strategy on the Flickr15 K data sets. The main contribution of our work is the preprocessing based on the boundary probability detector and the Chamfer distance transform and proposing a novel bag of mid maps descriptor. Results show that the proposal achieves significant improvements.