China
Africa
Explore chapters and articles related to this topic
Intelligent Machine Vision Technique for Disease Detection through Eye Scanning
Published in Biju Issac, Nauman Israr, Case Studies in Intelligent Computing, 2014
2.1 Introduction is chapter discusses a case study that involved a noninvasive and instant disease detection technique based upon machine vision through the scanning of the eyes of the subjects. e detected diseases involved conjunctivitis (eye u) and jaundice. In the proposed technique, color images of the sclera region of the subjects’ eyes were acquired by using an imaging setup specically designed for the purpose. e image acquisition setup consists of three separate charge-coupled device (3CCD) digital color camera kept in an aphotic enclosure. e facial region of the patient was lit under controlled illumination, and the images of the targeted region, namely, sclera region of the eyes, were acquired and processed using the algorithms developed in a MATLAB® environment to get the respective color attributes in RGB and La*b* color models. Principal component analysis (PCA)-based discrimination was applied over the color data of the subjects that showed high variance. e results of PCA indicated correlations among patients and the color attributes. e neuro-fuzzy-based software was developed for the prediction of jaundice and conjunctivitis along with the degree of severity and types, respectively. e experimental results showed good performance for the proposed method as compared to the conventional chemical methods with an accuracy level of over 90%.
Deep Learning for Retinal Analysis
Published in Ervin Sejdić, Tiago H. Falk, Signal Processing and Machine Learning for Biomedical Big Data, 2018
Henry A. Leopold, John S. Zelek, Vasudevan Lakshminarayanan
The Digital Retinal Images for Vessel Extraction database is a standardized set of fundus images commonly used to gauge the effectiveness of classification algorithms. The images are 8 bits per red green blue alpha (RGBA) channel with a 565 × 584 pixel resolution. The data set comprises 20 training images with manually delineated performance masks and 20 test images with two sets of manually delineated performance masks by the first and second human observers. The images were collected for a DR screening program in the Netherlands using a Canon CR5 nonmydriatic three charge coupled device (3CCD) camera with a 45-degree field of view [66].2
Identification of retinal diseases based on retinal blood vessel segmentation using Dagum PDF and feature-based machine learning
Published in The Imaging Science Journal, 2023
K. Susheel Kumar, Nagendra Pratap Singh
A 3CCD fundus camera captures the images, and each image from DRIVE has a dimension . The acquired images are captured with a 35-degree field view of the TopCon TRV-60 fundus camera. The images collected from the STARE dataset have the dimension of . The inputted RGB image is pre-processed through the PCA approach. The toggle contrast operator (TCO) enhances the resulting image from PCA. The image obtained from the pre-processing step for image samples are given in Figures 8 and 9 for different image samples. The image collected from the dataset is shown in Figure 8(a) and Figure 9(a). Next, the PCA is applied to transform the colour image into GSI, as shown in Figure 8(b) and Figure 9(b). The resulting image after applying CLAHE is shown in Figure 8(c) and Figure 9(c). These results show that the GSI was enhanced by the CLAHE approach but needs to be improved for efficient segmentation. Thus, TCO is applied to enhance the image segmentation, and the results of applying TCO are shown in Figure 8(d) and Figure 9(d).
Experimental and Computational Investigations of a Comb-Like Film-Cooling Scheme
Published in Heat Transfer Engineering, 2022
Hao-Ming Li, Wahid Ghaly, Ibrahim Hassan
The comb scheme geometry was manufactured on a test module, installed on the test plate, which is the bottom of the test section. The lateral width of blind slot is 63.5 mm (12.7 D). Downstream of the test module is a piece of extruded Polystyrene, covered with a TLC sheet. The size of the TLC sheet is 203 mm × 89 mm (40.6 D × 17.8 D). To decrease the end effect, only the central 25% was post-processed. A 3CCD camera is located right above the test section, to capture the TLC images through the transparent top of the test section. Each test records 300 TLC images at 5 frames/s. They were saved to a workstation with the simultaneous temperature data of the mainstream and the coolant, which were measured with thermocouples. The temperature uncertainty is ± 0.2 K. An in-situ TLC calibration was performed, in which calibrated temperature ranged from 22.1 °C to 28.3 °C. The post-processing used all these data with in-house programs to calculate η. The identical experimental system has accomplished massive projects, including [31–33], to list a few.
Experimental Study of Heat Transfer and Pressure Loss in Channels with Miniature V Rib-Dimple Hybrid Structure
Published in Heat Transfer Engineering, 2020
Figure 3 shows a schematic of the transient liquid crystal thermography for the present heat transfer experiments. The black painting was first sprayed directly onto the V rib-dimple test surface to improve thermochromic liquid crystal (TLC) color visibility for the image acquisition and then covered with a coating of TLC. The narrow bandwidth TLC (SPN100R35C1W from Hallcrest Ltd.) was used in the transient heat transfer experiments. The supplied TLC has a nominal red start temperature of 35 °C and a bandwidth of 1.0 °C. The TLC color changing process was recorded using a 3CCD video camera with a frame rate of 25 frames per second. When a hot air stream flows across the test surface coated with the TLC, the TLC starts changing the color depending on the temperature from red (about 35 °C) to blue (about 36 °C). The corresponding temperature-color relationship of the TLC was previously in situ calibrated in a steady state experiment with electrically heated copper plate using the same lighting and viewing conditions.