Image classification method based on hierarchical SIFT (scale-invariant feature transform) features and sparse coding

Abstract

The invention discloses an image classification method based on hierarchical SIFT (scale-invariant feature transform) features and sparse coding. The method includes the implementation steps: (1) extracting 512-dimension scale unchanged SIFT features from each image in a data set according to 8-pixel step length and 32X32 pixel blocks; (2) applying a space maximization pool method to the SIFT features of each image block so that a 168-dimension vector y is obtained; (3) selecting several blocks from all 32X32 image blocks in the data set randomly and training a dictionary D by the aid of a K-singular value decomposition method; (4) as for the vectors y of all blocks in each image, performing sparse representation for the dictionary D; (5) applying the method in the step (2) for all sparse representations of each image so that feature representations of the whole image are obtained; and (6) inputting the feature representations of the images into a linear SVM (support vector machine) classifier so that classification results of the images are obtained. The image classification method has the advantages of capabilities of capturing local image structured information and removing image low-level feature redundancy and can be used for target identification.

Description

technical field [0001] The invention belongs to the technical field of image processing, and relates to an image classification method, which can be used for object or target identification, and thus can be used for target detection and tracking. Background technique [0002] Image classification and recognition has always been one of the key research points in the field of image processing and computer vision. At present, the research on target recognition and image classification is booming at home and abroad. Target recognition is widely used in image panorama production, image watermarking, and robot globalization. Positioning, face detection, optical character recognition, manufacturing quality control, content-based image retrieval, target counting and monitoring, automatic vehicle parking systems, visual positioning and tracking, and video de-shaking, etc. The quality of image classification and recognition largely depends on how to effectively represent the image. I...

AI Technical Summary

Problems solved by technology

The method based on the underlying features currently exists mainly in combination with a single-layer network. For example, the linear spatial pyramid matching method based on sparse coding (Linear Spatial Pyramid Matching usingSparse Coding) proposed by J.Yang and proposed by A.Coates and A.Ng The disadvantages of the soft threshold voting method are: (1) the underlying features are redundant, and features that are not streamlined will increase the ambiguity of the later image classification; (2) directly using the underlying features loses local Structural information of image blocks; (3) A single-layer network cannot effectively simulate the human visual attention mechanism

Image pixel-based methods, such as Hierarchical Matching Pursuit (Hierarchical Matching Pursuit) proposed by L.Bo and Hierarchical Sparse Coding (Hierarchical SparseCoding) proposed by K.Yu, have the most significant disadvantage of not having local image block structure information

[0005] To sum up, the shortcomings of the above method are: on the one hand, the original image features are not simplified, resulting in that the redundant information of the extracted image underlying features is not removed, and the original image features do not capture the structural structure of local image blocks. information; on the other hand, the information represented by the image features constructed by a single-layer network has no depth, and it cannot describe the local prominent details in the image well, and cannot simulate the deep mining of the image after seeing an image. The ability of key information, such as locating the part with strong contrast or prominent edge in the image

Due to the above reasons, the accuracy of final image classification is relatively low

Images