Image Augmentation Using Hybrid RANSAC Algorithm

Abstract

The process of augmenting the number of images in a dataset is called Image Augmentation. Data volume is essential to process and generate digital outputs from a variety of features. This work focuses on the image augmentation using a hybrid RANSAC algorithm. The features extracted is used to join or merge the images by the blending of images. The proposed RANSAC algorithm is used to extract features from four images and produce the desired mosaiced image. A mosaiced picture is best suited for aerial photos and real-world objects. The blur metric of the proposed method is 185.2587 and which is 2.86% higher than the feathering blending algorithms. The total number of images in the dataset is 2100. The number of images after augmentation is 6300 with average accuracy of 95.6%. The reported remarkable results are beneficial to all the stakeholders on image augmentation.

ADD: An automatic desensitization fisheye dataset for autonomous driving

“…These enhancements are more generic in nature and are particularly relevant to expressing data transformation invariance. In addition, hybrid image augmentation [18]- [20] can mix cross-image information, which usually applies appropriate changes with labels to increase diversity. Furthermore, the adaptations of mixups [20], [49], [50] are popular among hybrid image augmentation.…”

“…To reduce annotation costs and further broaden the diversity of pedestrians, we further propose two data augmentation techniques: Occ-Data-Augmentation (ODA) and Pos-Data-Augmentation (PDA) for pedestrians' occlusions and postures, respectively. Using ODA and PDA, high-quality synthetic images are generated through the collecting, localization, and local fusion procedures, complementing the commonly used hybrid augmentation methods [18]- [20]. Besides, we build pedestrian detection baselines on our PPD dataset and extensive experiments validate the effectiveness of our novel data augmentation approaches.…”

RANSAC for Robotic Applications: A Survey

“…Image augmentation is the process of generating images similar to those present in a dataset, through several transformations, often with the aim of providing more training data to machine learning algorithms [ 79 ]. In [ 80 ], the authors propose a hybrid RANSAC algorithm to create a mosaic from several single images. They take images from similar areas and perform feature matching using RANSAC, using the location of those features to blend the pictures to create new ones.…”