A Superpixel Coding Image Retrieval Method Based on Inter-image Content Constraints

Abstract

The invention belongs to the technical field of image processing, and specifically discloses a superpixel coded image retrieval method based on content constraints between images. The steps are: 1) Carry out superpixel segmentation to the image, and record the position information of each superpixel block; 2) Extract the SIFT and LBP fusion features of the superpixel block of the image; 3) Randomly select the superpixel of the training image from the image database Block fusion features, generate a dictionary through K-means clustering; 4) Input the retrieval image, encode the retrieval image through the dictionary, and perform initial coding on the image except the retrieval image. 5) Perform initial encoding on the images in the image library, and perform constraint encoding on the images in the image library after screening the most similar encoding values ​​in the corresponding regions of the images to be retrieved as predictive encoding; 6) The images to be retrieved and the images in the image library The rest of the images are encoded, and the similarity matching is calculated, and the retrieval results are displayed according to the matching value. The present invention has higher accuracy rate and callback rate.

Description

technical field [0001] The invention belongs to the technical field of image processing, and in particular relates to a superpixel coded image retrieval method based on content constraints between images, which can be applied to human-computer interaction, information retrieval, and the like. Background technique [0002] Image retrieval has always been one of the key technologies that have attracted much attention in the field of computer vision, and it is also a research hotspot in the academic circles and major Internet companies. Although there are a large number of algorithms to achieve image retrieval, but because people do not have a unified standard for image annotation and understanding, with the improvement of the accuracy of people's information needs, the efficiency and accuracy of image retrieval also make image retrieval more difficult. development is difficult to meet people's needs. [0003] In recent years, superpixel segmentation has gradually become a pop...

AI Technical Summary

Problems solved by technology

It is difficult to achieve better retrieval results when the color of the image changes greatly

[0006] The above two methods are to extract feature codes from the image to be retrieved and the image in the image library, and fail to use the information of the image to be retrieved and the image in the image library. When the color characteristics and spatial position of the image change greatly, it is difficult to achieve efficient retrieval.

Images