Image retrieval method based on semi-supervised hashing

Abstract

The invention discloses an image retrieval method based on semi-supervised hashing, which mainly solves the problems of slow retrieval speed, large memory space occupation and inaccurate retrieval results in the prior art. The implementation steps are: (1) Extract the underlying features of the original image and normalize them; (3) Divide the normalized data into training data and test data; (4) Use the class label transfer method to obtain training data. The class label of the data, and the code is generated according to the class label; (5) use the training data and its code to train the support vector machine classifier; (6) use the support vector machine classifier to classify the training data and the test data, and generate the training data according to the classification results (7) Obtain the retrieval result according to the Hamming distance between the coding of the training data and the coding of the testing data. The invention reduces memory consumption, saves retrieval time, improves the accuracy of image retrieval, and can be used for e-commerce and image search services of mobile terminal equipment.

Description

technical field [0001] The invention belongs to the field of image processing, and further relates to an image retrieval method, which can effectively code the image and improve the retrieval precision of the image. Background technique [0002] With the rapid development of computer technology, digital multimedia technology and Internet technology, people's demand for information retrieval is also increasing. As a carrier of information, images contain rich content and bring great convenience to people's production and life. How to quickly and effectively search for the desired image has become one of the current research hotspots. In order to quickly and efficiently retrieve the images needed by users in big data, people encode the original images and use a certain length of hash code to represent the images. Since the storage and processing of data in the computer are hash codes, the use of hash codes can greatly speed up information retrieval. [0003] Tiange Technolo...

AI Technical Summary

Problems solved by technology

The disadvantages of this method are: the locality sensitive hashing method is a data-independent hashing method with strong randomness, and in order to ensure better retrieval accuracy, the number of coding bits required is very long, so it is difficult to implement in practical applications. very restrictive

However, the disadvantage of the method proposed in this patent application is that this method still does not avoid the premise that the training data in the spectral hash model is forced to obey a uniform distribution, which limits its application value

The shortcomings of this method are: the premise of this method is that the underlying features obey the Gaussian distribution, but the actual data may not obey the Gaussian distribution, and this method does not use the class label of the data, which reduces the retrieval accuracy

Images