A robust method for image stitching

Pellikka, Matti; Lahtinen, Valtteri

doi:10.1007/s10044-021-01005-8

Cited by 9 publications

(9 citation statements)

References 23 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Generally, this procedure is thought to provide better quality images compared to techniques that have only one image to leverage for matching features and edges to stitch individual adjacent fields of view. 29…”

Section: Methodsmentioning

confidence: 99%

Multicenter evaluation of the Vetscan Imagyst system using Ocus 40 and EasyScan One scanners to detect gastrointestinal parasites in feces of dogs and cats

Nagamori,

Scimeca,

Hall-Sedlak

et al. 2023

J VET Diagn Invest

View full text Add to dashboard Cite

The Vetscan Imagyst system (Zoetis) is a novel, artificial intelligence–driven detection tool that can assist veterinarians in the identification of enteric parasites in dogs and cats. This system consists of a sample preparation device, an automated digital microscope scanner, and a deep-learning algorithm. The EasyScan One scanner (Motic) has had good diagnostic performance compared with manual examinations by experts; however, there are drawbacks when used in veterinary practices in which space for equipment is often limited. To improve the usability of this system, we evaluated an additional scanner, the Ocus 40 (Grundium). Our objectives were to 1) qualitatively evaluate the performance of the Vetscan Imagyst system with the Ocus 40 scanner for identifying Ancylostoma, Toxocara, and Trichuris eggs, Cystoisospora oocysts, and Giardia cysts in canine and feline fecal samples, and 2) expand the assessment of the performance of the Vetscan Imagyst system paired with either the Ocus 40 or EasyScan One scanner to include a larger dataset of 2,191 fecal samples obtained from 4 geographic regions of the United States. When tested with 852 canine and feline fecal samples collected from different geographic regions, the performance of the Vetscan Imagyst system combined with the Ocus 40 scanner was correlated closely with manual evaluations by experts. Sensitivities were 80.0‒97.0% and specificities were 93.7‒100.0% across the targeted parasites. When tested with 1,339 fecal samples, the Vetscan Imagyst system paired with the EasyScan One scanner successfully identified the targeted parasite stages; sensitivities were 73.6‒96.4% and specificities were 79.7‒100.0%.

show abstract

Section: Methodsmentioning

confidence: 99%

Multicenter evaluation of the Vetscan Imagyst system using Ocus 40 and EasyScan One scanners to detect gastrointestinal parasites in feces of dogs and cats

Nagamori,

Scimeca,

Hall-Sedlak

et al. 2023

J VET Diagn Invest

View full text Add to dashboard Cite

show abstract

“…To handle low-quality and few-feature medical images, Chen et al [33] proposed an intensity-mosaic for automatic panoramic stitching of disordered images with insufficient features. Pellikka et al [34] proposed a robust method against repetitive patterns and featureless regions. Many other works [35,36] used deep learning for image stitching.…”

Section: Related Workmentioning

confidence: 99%

“…Calculating all local information from the global view can find the optimal global registration. The proposed method is inspired by [34] to construct a multigraph. The global information is represented by a multigraph, which is constructed using all the stitched images as nodes and registration relations as edges to connect adjacent nodes.…”

Section: Multigraph Constructionmentioning

confidence: 99%

Efficient stitching method of tiled scanned microelectronic images

Zhu

Han

Tao

2022

Meas. Sci. Technol.

View full text Add to dashboard Cite

Sparse features and repetitive textures are frequently presented in microelectronic microscopic images. Therefore, it is challenging for image stitching to meet the requirements of high-speed precision manufacturing. A novel image stitching method for tiled images is proposed to generate panoramic images of microelectronics quickly and accurately. According to the preset scan trajectory, grids were established between adjacent images for feature matching. The clustering algorithm was used to screen reasonable and multiple sets of registrations. Then, all registrations were used as connecting edges, and images were used as nodes, to create a multigraph. The unique registration in multigraph was solved by a non-linear minimization problem with linear constraints. Finally, image transformations were computed in global optimization for rendering panoramic images via image warping. The experimental results show that the proposed method improves the stability and efficiency of image stitching, furthermore, it maintains an equivalent level of precision as the Fiji and microscopy image stitching tool (MIST) methods.

show abstract

“…Using this method to extract the features and feature points of the image, the collected images require overlapping parts of adjacent pictures, which can ensure the accuracy of stitching. [47][48][49][50][51][52][53][54][55][56] SURF (SpeedUp Features) algorithm is mainly composed of three parts: feature point extraction and matching, image registration, and image synthesis. The transformation matrix H obtained during image registration is the core, The SURF algorithm is used to select and calculate the matching between the feature points to obtain the initial value of the transformation matrix between images.…”

Section: Depth Matrix Processing Based On Image Stitchingmentioning

confidence: 99%

“…Use a method based on fast and robust feature points to stitch images in the computer, compared with other splicing technologies, this technology has the characteristics of good robustness, good adaptability to image illumination changes, scale changes, and rotation changes, and the splicing speed is also fast. Using this method to extract the features and feature points of the image, the collected images require overlapping parts of adjacent pictures, which can ensure the accuracy of stitching 47‐56 …”

Section: System Designmentioning

confidence: 99%

Acquisition method of target leveling height based on machine vision

Yin

Song

et al. 2022

Concurrency and Computation

View full text Add to dashboard Cite

Summary Grain leveling robot is a special equipment to realize grain surface leveling in granary. Whether the target height of grain leveling is clear or not has a great influence on the leveling effect and efficiency of grain surface. In this article, the deficiencies of existing grain leveling robots are summarized, and a method of obtaining target leveling height based on machine vision is proposed. Data were collected by RGB‐D camera, the depth matrix of the whole large granary was obtained by image stitching and edge detection, and then the target height of the grain leveling robot was obtained by the element method. It solves the problem of controlling the falling height of the robot by human subjective feeling, improves the efficiency and precision of the robot, optimizes the effect of the robot, and helps to realize the intelligent work of the robot.

show abstract

A robust method for image stitching

Cited by 9 publications

References 23 publications

Multicenter evaluation of the Vetscan Imagyst system using Ocus 40 and EasyScan One scanners to detect gastrointestinal parasites in feces of dogs and cats

Multicenter evaluation of the Vetscan Imagyst system using Ocus 40 and EasyScan One scanners to detect gastrointestinal parasites in feces of dogs and cats

Efficient stitching method of tiled scanned microelectronic images

Acquisition method of target leveling height based on machine vision

Contact Info

Product

Resources

About