Study On Coding Of Elemental Image Array

As the most promising true three-dimensional technology,integral imaging technology has been widely used in many fields,and related technology researches are developing,including the acquisition and display of integral imaging system,as well as the coding of the elemental image array discussed in this paper.The image displayed by the integral imaging system is an elemental image array.More and more large scale actual shooting scene images need to be converted into high resolution elemental image array for integral imaging display,so this leads to more demands on transmission bandwidth and compressed storage space.Therefore,it is of great significance to study the coding method of elemental image array with a large amount of data.At present,most researches on the coding of elemental image arrays at home and abroad are based on the correlation of the elemental image arrays,but the correlation is searched and coded in different ways.Among them,the most coding methods studied are based on the international standards of video coding such as H.264 and H.265,together with the non-standard methods based on wavelet transform and three-dimensional Discrete Cosine Transformation(3D-DCT).These methods do not explore the correlation of the elemental image array from the source,thus limiting its coding performance.In this paper,the feature of the elemental image array is distinguished from the ordinary two-dimensional image.The correlation between the adjacent elemental images in the elemental image array is accurately calculated by the physical parameters of the lenses array during the acquisition of the elemental image array.Two coding methods proposed for elemental image arrays are as follows:(1)The first is the coding method which utilizes the one-dimensional correlation of elemental image array.The theoretical basis of this method is to combine the imaging geometric features of the elemental image array,and calculate the offset of the homologous image points between adjacent elemental images in the horizontal direction through the relevant physical parameters of the lenses array.Then we obtain the area of the overlap area between adjacent elemental images in the horizontal direction.In this paper,the elemental images in the horizontal direction of the elemental image array are grouped according to the offset values of the homologousimage points,and at the same time,specific coding order of each elemental image in each group is determined.After that,we obtain the prediction images of the elemental images to be coded according to the pixel values of the homologous image points between adjacent coded elemental images.Finally,the residual images are coded.(2)The second is the coding method which uses the two-dimensional correlation of elemental image array.The theoretical basis is the offset of the homologous image points between elemental images in the elemental image array.All original reference elemental images are determined by calculating the offset of the homologous image points between adjacent elemental images in the horizontal and vertical directions,which are first encoded.Then,the coding order and the prediction image of the other remaining elemental images are determined using the two-dimensional correlation of the elemental image array.The pixel values of the prediction images are the pixel values of the homologous image points between adjacent encoded elemental images in the horizontal or vertical direction,and finally the residual images are also encoded.In the end,the images with different sizes and different scenes are used to verify the two coding methods under High Efficiency Video Coding(HEVC)platform,compared with the classic HEVC algorithm.In order to optimize the results of the HEVC algorithm when coding using the HEVC classical algorithm,all the elemental images in the elemental image array are scanned into a sequence of video streams in the order of Z words as input images.The experimental results show that the two coding methods proposed in this paper have a larger performance improvement than the HEVC algorithm,especially the coding method based on the two-dimensional correlation of elemental image array,its performance is partially improved compared with the method based on the one-dimensional correlation of elemental image array.The two coding methods proposed in this paper reduce not only the coding rate but also the complexity of the algorithm under the premise of ensuring the reconstructed image quality.