The Image Interpolation Method Based On Image Low-rank And The Implementaion Based On Hadoop

In recent years, one of the branches of image processing is about image restoration. Image interpolation technology is a very important kind of image restoration method. It has a very wide range of applications in many aspects of life, such as medical imaging, satellite remote sensing,military radar and astronomical observation.This paper researches the method of recovering the image which is captured by the sensor with the Bayer color filter array (CFA) in the digital camera. In order to reduce the volume of the digital cameras and other digital image acquisition systems and to save their cost, digital cameras and video cameras usually have only one sensor, and this sensor only need to install a color filter array in front of it, the image obtained in this way belongs to low resolution image, also it can be called mosaicked images. Mosaicked images can be restored into the images with complete color components through image interpolation technology and the restored images can be seen as high resolution images. The low resolution image also saves the storage space in the camera, so this technology has been widely used in the digital image acquisition systems.This paper proposes an image interpolation method based on the properties of low rank image.This method uses the low rank property, constructs an optimization objective function, and through the augmented Lagrange function method and alternating direction method to do the iteration and to do the interpolation to mosaicked image. The matrix based on the image is high dimensional matrix.the most parts of high-dimensional data in the image can be expressed as the sum of a low rank matrix and noise. Generally speaking, a lot of low rank matrix problems can be solved through the optimization of the objective function, and uses convex relaxation of objective function to replace the original function. To use convex optimization problem to solve convex relaxation is in order to find the optimal solution.This paper also proposes a method which uses the Hadoop platform to achieve two traditional image interpolation methods and the method based on the properties of low rank image. The traditional methods are the bilinear interpolation method and edge-directed interpolation method.Because interpolation technology sometimes needs to be real-time, but many image interpolation technologies are too long to meet the demand. So this method uses distributed cluster system of Hadoop platform, divides the image into many small images, and uses multiple clusters to do the interpolation to small images at the same time. This can improve operation efficiency and reduce operation time.