Hand-drawn sketch retrieval method based on deformable convolution and depth network

Abstract

The invention belongs to the field of computer vision and depth learning, the invention particularly discloses a hand-drawn sketch retrieval method based on deformable convolution and depth network, The method comprises the following steps: S1, acquiring a hand-drawn sketch and natural color map database S2, A natural color image is converted into an edge map S3 by an edge detection algorithm, Thedepth network S5 based on deformable convolution is trained by preprocessing S4 of the hand-drawn sketch and the edge map through morphological operation, and the depth features S6 of the hand-drawnsketch and the edge map of the natural image are extracted respectively by using the trained depth network, and similarity between the features is calculated and the retrieval result is returned. Themethod of the invention has the beneficial effects that deformable convolution is incorporated into the traditional neural network, the limitation of the standard convolution on the hand-drawn sketchcan be broken, the robustness of extracting features of the network hand-drawn image can be improved, and the feature redundancy can be reduced. The network structure proposed by the invention can greatly improve the retrieval precision of the hand-drawn sketch.

Description

technical field [0001] The invention belongs to the field of computer vision and deep learning, and relates to a hand-drawn sketch retrieval method based on deformable convolution and deep network. Background technique [0002] Hand-drawn sketch retrieval is a kind of content-based image retrieval technology. Studies have shown that hand-drawn pictures have the same mechanism as real pictures to activate the visual area of ​​human cerebral cortex. Compared with text-based image retrieval and traditional image retrieval based on natural color maps, hand-drawing has the advantages of easy acquisition, strong abstraction and generalization, and no language and cultural restrictions. The number of touch-screen devices is increasing day by day, and the acquisition of hand-drawn images is becoming easier. Image retrieval based on hand-drawn sketches is currently receiving more and more attention and has broad application prospects. For example, commercial online shopping malls us...

AI Technical Summary

Problems solved by technology

When using deep neural networks to extract features, the current mainstream neural networks use regular convolution for feature extraction. Regular convolution is very effective for information-intensive natural images, but for hand-drawn sketches, a large number of useless features will be extracted and there are Severely redundant, i.e., traditional neural network structures are not well suited for hand-drawn sketches

Images