Image Reconstruction Research Based On Electrostatic Field Model

Digital image reconstruction is an important research in the field of image processing and aresearch hotspot in computer vision and computer graphics. It can be used to reconstruct image’sscratches or missing parts and remove the text or other specific goals in the images, in order toachieve the determined target of reconstruction or realize the preprocessing of in-depth analysis ofthe images. The present study of image reconstruction does not aim at being detected by humaneyes. It focuses on the overall visual effect, not the pursuit of the reconstruction’s accuracy in aspecific area of the image. However, with the widely application of images in image analysis andrecognition, the reconstructed images are more frequently used for detail information extraction.Therefore, how to accurately reconstruct the missing information of the image has caused theattention of researchers.At the same time, more and more videos with the real-time requirementneeds to be processed. Therefore the problem of reconstruction efficiency has been considered.The image has an inherent relationship with the stability field as the image is the stability resultof the interaction between the object surface texture and the structure with the imaging light. Basedon local texture stability field model, the electrostatic field model is selected for the image’s localarea texture modeling. The image’s local area reconstruction model is proposed based on pointsources influence function. This model can achieve the purpose of accurate reconstruction of thedamaged pixel point by calculating the around known points’ influence on this point. Finally,according to the electrostatic field’s potential derivation, the constraints are analyzed and acalculation method of point source influence function is determined to achieve a modelreconstruction. Experimental results show that the proposed not just can keep a relative ideal visualeffect, but also have a relative higher efficiency and accuracy.In the above reconstruction process of the electrostatic field model, the choice of point sourceinfluence function will affect the reconstruction effect. In this paper, the broken edge is divided intoa single value edge, multi-value edge and blend edge by analyzing the type of edge damage. Theexpression of the point source influence function is adjusted based on the type of edge. And whencalculating the multi-value edge, a certain size of the reconstruction area is selected for damagedpixels in multi-value edge. The pixel values of the reconstructed area are screened by direction toget the main direction gradient of the reconstruction region, therefore the reconstruction process isstrictly performed according to the direction of isophote. This result will perform better on edgeconnectivity. Experimental results show that this algorithm makes full use of effective information around the broken edges. The broken image has better edge connectivity and the visual effect isenhanced after reconstructing.