OCPAD — Occluded checkerboard pattern detector

Abstract

Many camera calibration techniques require the detection of a pattern with known geometry, e.g., a checkerboard. Typically, the pattern must be fully contained in the field of view. This brings several limitations, one of which is that lens distortion can not reliably be estimated in outer image regions.This paper presents the occluded checkerboard pattern detector (OCPAD) to find checkerboards, even in a) lowresolution images, b) images with high lens distortion and if c) the pattern is partly occluded or not completely within the field of view. We exploit that checkerboards can easily be represented by a graph. We use graph matching to find the largest partial checkerboard in the image. Our detector complements a state-of-the-art calibration algorithm. Quantitatively, detection rates are considerably improved over the state-of-the-art. Additionally, estimation of lens distortion is greatly improved at outer image regions. Here, the reprojection error is improved by up to 50%.

Automatic checkerboard detection for camera calibration using self-correlation

“…Some other automatic checkerboard detection methods and implementations are also applied in the experiments for comparison, including the method using a single shot proposed by Geiger et al [12], the Camera Calibrator app included in the '2016b' version of MATLAB [23], the ROCHADE method [9] and the OCPAD method [10]. The quantitative detection results of the methods are shown in Table 3 (bold values are best results of each dataset).…”

“…Placht et al used an edge graph generation idea to accurately refine the checkerboard corners and the algorithm can process images at extreme poses or with high distortions, and it is also valid for low-resolution images [9]. Fuersattel et al then made an improvement that uses graphs to represent checkerboards and can find partly occluded checkerboard pattern by graph matching, resulting in a higher detection rate and a shorter runtime [10]. Bok et al employed circular boundaries of corner candidates to extract the checkerboard ones and conduct the indexing, which also performs well in processing distorted or noisy images [11].…”

“…Our approach also contains a checkerboard structure recovery solution using gradient, orientation and distance constraints of the checkerboard edges, which is able to detect multiple checkerboards in a single image through structure expansion and grouping. We evaluate the proposed approach with six datasets including five online public datasets [9,10,14], and compare it to the state-of-the art methods. The main contributions of the proposed method are: (1) detecting checkerboard corners using self-correlation; (2) recovering the structure of a checkerboard by taking the advantages of the grey distribution in the checkerboard square as well as angle and distance constraints; (3) gaining robust results against distortion with a high detection rate.…”

“…The first three, the Mesa SR4000 (176×144 pixels), the IDS uEye (1280×1024 pixels) and the GoPro Hero 3 (4000×3000 pixels) datasets, are from the evaluation data of [9]. The other two public ones are the Partial-full datasets (1280×720 pixels) from [10] and the images (1392×512 pixels) from the KITTI calibration data [14]. The most significant difference between the KITTI calibration data and the other four datasets is that there are more than ten checkerboards in each image of the KITTI data while for the others there is only a single one.…”

Segmentation of photovoltaic module cells in uncalibrated electroluminescence images

“…The segmentation of PV modules into individual solar cells is related to the detection of calibration patterns, such as checkerboard patterns commonly used for calibrating intrinsic camera and lens parameters [29,36,41,69,79]. However, the appearance of calibration patterns is typically perfectly known, whereas detection of solar cells is encumbered by various defects that are a priori unknown.…”