Depth range detection by producing object-based blurring effect

摘要

This paper presents a new image segmentation method for detecting depth range of object using two images focused on two different depth. Assume the image scene consists of some layers, linear imaging model is introduced to analize defocus parameter in each layer of depth. Based on the model, three images with different blurring effect are generated by linear filtering of the acquired images for emphasizing the defference of defocus parameter of each depth. Cost function is evaluated and thresholded compared to theoretical value for detecting the region corresponding to the depth using DFT of sub-blocks of their generated images: Simulation and experimental results for real images are shown to test the performance of the proposed method.