Depth estimation from single image using Defocus and Texture cues

Abstract

As imaging is a process of 2D projection of a 3D scene, the depth information is lost at the time of image capture from conventional camera. This depth information can be inferred back from a set of visual cues present in the image. In this work, we present a model that combines two monocular depth cues namely Texture and Defocus. Depth is related to the spatial extent of the defocus blur by assuming that more an object is blurred, the farther it is from the camera. At first, we estimate the amount of defocus blur present at edge pixels of an image. This is referred as the Sparse Defocus map. Using the sparse defocus map we generate the full defocus map. However such defocus maps always contain hole regions and ambiguity in depth. To handle this problem an additional depth cue, in our case texture has been integrated to generate better defocus map. This integration mainly focuses on modifying the erroneous regions in defocus map by using the texture energy present at that region. The sparse defocus map is corrected using texture based rules. The hole regions, where there are no significant edges and texture are detected and corrected in sparse defocus map. We have used region wise propagation for better defocus map generation. The accuracy of full defocus map is increased with the region wise propagation.

Blur parameter locus curve and its applications

“…Zhuo and Sim [9] produce reliable defocus map in the presence of noise employing gradient ratio at the edges. Srikakulapu et al [51] use reliable local scale using Elder and Zucker's method [43] to obtain the edge points at which defocus is estimated. However, the presence of non‐ideal edges, false edges and other types of blurs affect the performance of these techniques.…”

“…14. First column shows the test images, the second column shows the defocus map results for Zhuo's method [9], third column shows defocus map obtained using Srikakulapu's method [51] without applying the hole filling, fourth column shows defocus map for Karaali's method [52] by utilising edge‐aware matting given in [52], fifth column shows defocus map for Karaali's method [52] using Levin's [54] closed‐form matting and last column shows defocus map obtained with the proposed correction. Sky region in the third image introduces large error in estimated defocus map as no defocus information is present for such regions.…”

“…However, Karaali's method does not effectively remove edge locations corresponding to erroneous defocus values which in turn result in the erroneous defocus maps. Similarly, defocus maps obtained using Zhuo's method [9] and Srikakulapu's method [51] also contain erroneous regions. Thus, the proposed method results in more accurate and reliable defocus map creation compared to state‐of‐the‐art methods.…”

“… Defocus map comparison, Column (a) shows test images, Columns (b), (c), (d) and (e) show defocus maps obtained using Zhuo [9], Srikakulapu [51], Karaali 1 [52], Karaali 2 [52] and proposed methods respectively; Karaali 2 utilises edge‐aware matting given in [52] to obtain full defocus map from sparse defocus map, Karaali 2 utilises Levin's [54] closed‐form matting to obtain the full defocus map from the sparse defocus map (a) Image, (b) Zhuo [9], (c) Srikakulapu [51], (d) Karaali 1 [52], (e) Karaali 2 [52], (f) Proposed…”

Monocular 3D metric scale reconstruction using depth from defocus and image velocity

“…Third, the blur texture ambiguity is still a challenging problem. Srikakulapu et al [18] proposed a method to correct the depth map by using texture information such as edge sharpness, spot energy, and contrast. However, this approach cannot estimate the metric scale.…”