Blog column>Internet column>Image super resolution technology
  • Image super resolution technology
  • Image super resolution (Super Resolution Image) refers to a low resolution image or image sequence to restore a high resolution image. Image super resolution technology is divided into super resolution reconstruction and super resolution reconstruction. At present, the research of image super resolution can be divided into 3 main categories: interpolation based, reconstruction based and learning based methods.
  • subscribe

Updated articles

[programming language] Machine learning - deep learning (Learning Deep)

Deep learning is a machine learning in a very close to the AI field, the motivation is established, to simulate the human brain to analyze learning neural network, recently studied the knowledge of machine learning and deep learning, this paper gives some useful information and experiences. Words Key: supervised learning and unsupervised learning, classification, regression, density estimation, clustering, deep learning, DBN Sparse, 1 supervised learning and unsupervised learning Given a set of data (inpu...

[programming language] Machine learning combat notes -- handwriting recognition system based on KNN algorithm

Realization of handwriting recognition system using the k-nearest neighbor classifier, about 2000 samples of the training data set, each number about 200 samples, each sample preservation in a TXT file, handwritten body image itself is 32x32 binary image, as shown below: First, we need to format the image as a vector, the 32X32 binary image matrix through the img2vector () function is converted to 1X1024 vector: Img2vector def (filename): Retu...

[programming language] Machine learning combat notes - the use of KNN algorithm to improve the matching effect of dating sites

First, the case background My friend Helen has been using online dating sites to find the right date. Although the dating site will recommend different candidates, but she does not like everyone. After a summary, she found that she had had three types of people: (1) the person who does not like; (2) the charm of the general people; (3) the most attractive person; Although found in the law, but Helen still unable to take dating site recommended matching object placed in appropriate classification, she felt can be general in the Monday to Friday date the charm and weekend prefer and those very charming companionship...

[Internet] Machine learning real note - KNN algorithm

One, KNN algorithm description KNN (neighbor algorithm K-nearest), that is, K nearest neighbor algorithm, as its name suggests, can be the image of the understanding of the K nearest neighbors. When K=1, KNN algorithm has become the nearest neighbor algorithm, that is, to find the nearest neighbor. So-called k nearest neighbor algorithm is given a training data set of new input instances in training data concentration found with the example of the nearest K instances (that is, mentioned above K neighbors), if the K instances of majority belongs to a class, the input instance classification to the...

[programming language] "Machine learning" notes - SVD based image compression

The original image size is 32 32=1024 X pixels, the use of SVD to reduce the dimension of the data, to achieve image compression Create a new svdRec.py file and add the following code: #printMat () function for printing matrix PrintMat def (inMat, thresh=0.8): I in range for (32): K in range for (32): FL if...

[Internet] "Machine learning actual combat" notes: using SVD to simplify data

SVD (Value Decomposition Singular) singular value decomposition, can be used to simplify the data, remove the noise, improve the results of the algorithm. One, SVD and recommendation system The following figure from the restaurant dishes and food division of the views of these dishes, the food division can be used between 1 to 5 of an integer to the vegetable rating, if the food division did not taste the dish, the rating is 0 Create a new file svdRec.py and add the following code: LoadExData def (): Return[...

[programming language] Sparse representation of the dictionary (MATLAB implementation code)

The main purpose of this paper is to achieve the sparse representation based image super resolution of the image Super-Resolution via sparse representation "in Figure2. Through to 100000 high-resolution and low resolution image block train and get the high resolution image patch dictionary, dictionary of atoms for a total of 512 and atomic size is 9x9 One method: Clc; All clear; % dictionary load Load ('Diction...

Imshow OpenCV can not normally display pictures

Question: imshow OpenCV can not normally display the picture. Solution: (imshow) add a waitKey statement (30) Main int () { InputImg = imread ("input.bmp"); Imshow ("input image", inputImg); / / display image //waitKey (30); ImgSplit (inputImg); / / get the M (x) ImgBlur...

Compressive sensing -- Hadamar Walsh (WHT) transform and inverse transform of Matlab code implementation

Is a kind of typical non sine function transformation wht) Walsh Hadamard transform (Walsh-Hadmard transform. Using orthogonal function as the basis function, with properties similar to Fourier function, image data more is evenly distributed, after Walsh Hadamard transform data becomes more concentrated on the corner of the matrix, so Walsh transform with energy concentrated in nature, and can be used to compress the image information. Hadamard function in Matlab: Format: H=hadamard (n), returns a n * n...

Python achieve K-means clustering

Kmeans is one of the most simple clustering algorithm, but it is widely used. Recently, this algorithm is often encountered in the work. Kmeans general in the data analysis of the early use, select the appropriate K, the data classification, and then classification study the characteristics of different data clustering. Kmeans algorithm steps: 1 randomly selected k center point 2 traverse all data, divide each data into the nearest Center 3 to calculate the average value of each cluster, and as a new center point 4 repeat 2-3 until the K line point is no longer changed...

[programming language] LBP texture feature

Local binary pattern (local binary patterns LBP) the first is as a kind of effective texture description operator is proposed, due to the local texture feature of the image mapping excellence ability obtains the very extensive application. LBP feature has a strong classification ability (Highly Discriminative), higher computational efficiency and invariance to monotonic gray scale changes. The LBP method first in 1994 by T. Ojala, M.Pietik by inen, D. and Harwoo...

[comprehensive] Image super resolution based on sparse representation "Super-Resolution Via Sparse Representation Image"

Because recently doing image super resolution reconstruction research, lucky enough to see the Yang Jia teachers and Ma Yi teacher Daniel, published in 2010 article on image super resolution of classic essay "ImageSuper-Resolution "Sparse Representation Via", so the paper on the translation, such as improper, but also please help a lot of correction!!! Yang Jianchao, Wright John, Huang Thomas, and...

Image super resolution ImageSuperReso Sparse representation Over complete dictionary learning machine learning algorithm

  • Columnist:Geekmanong
  • Creation time: 2015-12-11
  • Article number: 13
  • Browse volume: 6455 times

Hot articles

Latest comments