Single sensor camera obtains the co

Single sensor camera obtains the color information of the image through the color filter array (CFA), but the collected image has only one original color component at each pixel position. In order to recover the full-color image obtained by the other two components, it must be interpolated. This process is also called color de mosaic. After the scene or object is sampled by Bayer color of single sensor digital camera, a mosaic image is obtained, that is, each pixel has only one color value. In order to get a full-color image, we need to carry out color interpolation, and estimate the other two missing color values at each pixel position, which is also the problem of de mosaic. The single sensor digital camera can only capture one third of the RGB main image color, and the sample data only accounts for one third of the whole data. The image is reconstructed from the full resolution column or from the incomplete data. This process is called color visualization. Similarly, digital cameras generally use CCD (charge coupled device) as the sensor to receive images. In the process of photographing, each pixel can only get one color, and the image is composed of three colors: red, blue and green. In order to get full color images, two other missing color components must be recovered. This process is called demosaicking. At present, most of the methods use the correlation of color channels and the local geometric properties of the image to interpolate to remove mosaic, which mainly produces two kinds of errors: sawtooth phenomenon and false color. The quality of image restoration plays an important role in the follow-up processing of computer vision system, such as image recognition, image segmentation and so on. Therefore, the study of effective De mosaic model and algorithm has important theoretical and application value.