To study the problem of real-time and accuracy of the image retrieval algorithm of embedded system with scale, rotation and illumination. Improved SURF algorithm, which is applied to the binary image feature extractio...
详细信息
ISBN:
(纸本)9781538612309
To study the problem of real-time and accuracy of the image retrieval algorithm of embedded system with scale, rotation and illumination. Improved SURF algorithm, which is applied to the binary image feature extraction, improve the real-time performance of the image, the use of LSH algorithm to establish the index of the image feature database, to avoid duplicate feature extraction, the use of LSH algorithm to approximate the search image features, and the similarity into line sort, select the best matching image. When the image has scale, rotation, illumination, this algorithm has a higher retrieval accuracy than the SURF algorithm and ORB algorithm, and has a better robustness than the SURF algorithm. The experiments show that the algorithm has scale, rotation and illumination conditions of the embedded system of imageretrieval, has good real-time performance and achieved good application effect.
This article expounds that color distribution entropy is used to describe the distribution characteristics of the spatial distribution of color in pictures, and makes a further improved algorithm according to human vi...
详细信息
This article expounds that color distribution entropy is used to describe the distribution characteristics of the spatial distribution of color in pictures, and makes a further improved algorithm according to human visual and entropy characteristics. We can show that the method of imageretrieval is effective and has higher retrieval performance by analyzing the experimental results.
This paper proposes a novel image retrieval algorithm based on wavelet packet histogram techniques. Firstly, we decompose an image with a family of real orthonormal wavelet bases and compute the energy using wavelet p...
详细信息
ISBN:
(纸本)0780393953
This paper proposes a novel image retrieval algorithm based on wavelet packet histogram techniques. Firstly, we decompose an image with a family of real orthonormal wavelet bases and compute the energy using wavelet packet coefficients. Secondly, we select few number of most dominant energy channels for thresholding and non-linear filtering. Finally, we compute the wavelet packet histogram as features and employ histogram intersection distance to retrieve queried image from image databases, the proposed method employs a smaller feature space and involves as less computation cost in the feature extraction. The experimental results show that these techniques can archive higher performance in the imageretrieval.
With the rapid development of network technology, the number of digital images is growing at an alarming rate, people's demand for information gradually shift from text into images. However, it is very difficult f...
详细信息
With the rapid development of network technology, the number of digital images is growing at an alarming rate, people's demand for information gradually shift from text into images. However, it is very difficult for users to quickly find the images they are interested in from the large number of image libraries. The purpose of this paper is to study the image recommendation algorithm based on deep learning. In this paper, image classification algorithm is firstly studied. LReLU - Softplus activation function is formed by combining LReLU function and Softplus function, and CNN is improved. Then, an imageretrieval model based on local sensitive hash algorithm is proposed in this paper. This model calculates the distance in hamming space for the binary hash code generated by mapping. Euclidean distance is calculated inside the result set after similarity measurement to improve the accuracy, and the imageretrieval model is constructed. Finally, an image recommendation model based on implicit support vector machine (SVM) is proposed in this paper. This image recommendation method combines image text information and image content information. The experimental results show that the proposed image recommendation model can meet the practical needs. In this paper, the overlap rate between the CNN-based recommendation model and the human recommendation algorithm was tested, and the coincidence degree of the two recommended images reached 88%.
Campus university library as a spiritual force, in the development of social economy and science and technology innovation, according to the requirements of education management research scholars and students growth c...
详细信息
Campus university library as a spiritual force, in the development of social economy and science and technology innovation, according to the requirements of education management research scholars and students growth changes proposed learning Shared space, the effective integration of space, resources, service as one of the new construction idea, not only makes the university library become more perfect effective, It can also enhance the cohesion and innovation of internal operation. Nowadays, with the increase of the number of university libraries around the country, the scale of internal construction is getting bigger and bigger, and the planning and design of learning shared space based on university libraries has become the focus of scientific research. In essence, learning shared space, as a seamless learning environment, begins to innovate toward intellectualization and modernization while transforming the functional role of university libraries. In this article, therefore, learn about the current university library, on the basis of Shared space planning and design content, according to the fuzzy matrix model, analysis of the application of image retrieval algorithm is verified, the final results show that in the university library to promote learning sharing space, can not only meet the diverse needs of the readers, you can also change the traditional mode of the library service.
暂无评论