Experiment of OCITN: Considering Appropriate Goal Images and Metric for One-Class Image Transformation Network

One-class classification (OCC) is a classification problem where training data includes only one class. In such a problem, two types of classes exist, seen class and unseen class, and classifying these classes is a challenge. Besides, One-class Image Transformation Network (OCITN) is an OCC algorithm for image data. In which, image transformation network (ITN) is trained. ITN aims to transform all input image into one image, namely goal image. Moreover, the model error of ITN is computed as a distance metric between ITN output and a goal image. Besides, OCITN accuracy is related to goal image, and finding an appropriate goal image is challenging. In this paper, 234 goal images are experimented with in OCITN using the CIFAR10 dataset. Experiment results are analyzed with three image metrics: image entropy, similarity with seen images, and image derivatives.

Assessment of Suitable Image Fusion Method for CARTOSAT-2E Satellite Urban Imagery for Automatic Feature Extraction

The purpose of this study is to investigate the best suitable pan sharpening method for CARTOSAT-2E satellite launched by ISRO (Indian Space Research Organisation). This satellite provides high resolution images that are being used for many urban applications such as mapping, feature extraction, facility management etc. The synthesized image using pan sharpening method enables users to take the combined advantage of the best available spatial and spectral resolutions. In this paper, various pan sharpening methods based on component substitution (CS) and Multi Resolution Analysis (MRA) are applied on the CARTOSAT-2E images and the resultant images are tested for their qualitative and quantitative performance. Qualitative analysis is carried out based on image blur and spectral distortion and quantitative evaluation is performed using image metrics by comparing the synthesized image with the original image. The results show that the High-Pass Filter (HPF) method offers the good spectral fidelity. However, due to its inherent stair-casing effect in the resultant image; modified-IHS followed by PRACS method is found to be preferable for automatic urban feature extraction from CARTOSAT-2E images.

Enhanced Artificial Intelligence System for Diagnosing and Predicting Breast Cancer Using Deep Learning

Breast cancer is the leading cause of death among women with cancer. Computer-aided diagnosis is an efficient method for assisting medical experts in early diagnosis, improving the chance of recovery. Employing artificial intelligence (AI) in the medical area is very crucial due to the sensitivity of this field. This means that the low accuracy of the classification methods used for cancer detection is a critical issue. This problem is accentuated when it comes to blurry mammogram images. In this paper, convolutional neural networks (CNNs) are employed to present the traditional convolutional neural network (TCNN) and supported convolutional neural network (SCNN) approaches. The TCNN and SCNN approaches contribute by overcoming the shift and scaling problems included in blurry mammogram images. In addition, the flipped rotation-based approach (FRbA) is proposed to enhance the accuracy of the prediction process (classification of the type of cancerous mass) by taking into account the different directions of the cancerous mass to extract effective features to form the map of the tumour. The proposed approaches are implemented on the MIAS medical dataset using 200 mammogram breast images. Compared to similar approaches based on KNN and RF, the proposed approaches show better performance in terms of accuracy, sensitivity, spasticity, precision, recall, time of performance, and quality of image metrics.

Defining the Visual Complexity of Learning Management Systems Using Image Metrics and Subjective Ratings

Research has demonstrated that students’ learning outcomes and motivation to learn are influenced by the visual design of learning technologies (e.g., learning management systems or LMS). One aspect of LMS design that has not been thoroughly investigated is visual complexity. In two experiments, postsecondary students rated the visual complexity of images of LMS after exposure durations of 50-500 ms. Perceptions of complexity were positively correlated across timed conditions and working memory capacity was associated with complexity ratings. Low-level image metrics were also found to predict perceptions of the LMS complexity. Results demonstrate the importance of the visual complexity of learning technologies and suggest that additional research on the impact of LMS design on learning outcomes is warranted.

Application-Specific Evaluation of a Weed-Detection Algorithm for Plant-Specific Spraying

Robotic plant-specific spraying can reduce herbicide usage in agriculture while minimizing labor costs and maximizing yield. Weed detection is a crucial step in automated weeding. Currently, weed detection algorithms are always evaluated at the image level, using conventional image metrics. However, these metrics do not consider the full pipeline connecting image acquisition to the site-specific operation of the spraying nozzles, which is vital for an accurate evaluation of the system. Therefore, we propose a novel application-specific image-evaluation method, which analyses the weed detections on the plant level and in the light of the spraying decision made by the robot. In this paper, a spraying robot is evaluated on three levels: (1) On image-level, using conventional image metrics, (2) on application-level, using our novel application-specific image-evaluation method, and (3) on field level, in which the weed-detection algorithm is implemented on an autonomous spraying robot and tested in the field. On image level, our detection system achieved a recall of 57% and a precision of 84%, which is a lower performance than detection systems reported in literature. However, integrated on an autonomous volunteer-potato sprayer-system we outperformed the state-of-the-art, effectively controlling 96% of the weeds while terminating only 3% of the crops. Using the application-level evaluation, an accurate indication of the field performance of the weed-detection algorithm prior to the field test was given and the type of errors produced by the spraying system was correctly predicted.

Estimation of the Ki-67 Proliferation Marker Expression Level with Computer Vision Technology

В последнее время в качестве средств компьютерной диагностики широкую популярность приобретают элементы из области компьютерного зрения. Целью данной работы является разработка программных средств, позволяющих автоматически анализировать онкомаркированные клетки в образцах медицинских изображений. В процессе работы была решена такая задача, как выявление количества онкомаркированных клеток в образце медицинского изображения. Это стало возможно благодаря тесному взаимодействию с морфологами, обозначавшими интересующие их фрагменты снимка, где необходимо было произвести подсчет онкомаркированных клеток, и опыту программиста в решениях подобных задач других областей жизнедеятельности с помощью технологии компьютерного зрения. Медицинские изображения были размечены морфологами и представляли собой срез стекла, на котором были окрашены в определенный цвет онкомаркированные клетки специальным препаратом. Исследуемым органом стал язык. В результате был совершен подсчет иммунопозитивных клеток, что дает морфологам возможность наблюдать более объективную картину в изображении среза стекла образца, позволяющую скорректировать решение в отношении постановки диагноза и назначения необходимого лечения. Lately, computer vision technologies have become common computer diagnostics tools. The study objective is the development of software tools to automatically detect cells with proliferation markers on medical images. We managed to qualitatively assess the number of cells with proliferation markers on a medical image sample. We closely cooperated with morphologists who identified the areas of interest on the image where the cells with proliferation markers were to be counted. Another enabler was an extensive experience with developing computer vision systems for other applications. The medical images were tokenized by the morphologists. The images were slices where the proliferation marker cells were dyed. The organ under investigation was the tongue. We counted the immune-positive cells. It helped the morphologists to have some objective slice image metrics used to adjust the diagnosis and therapy plans.

Multimodal Medical Image Fusion using NSCT and DWT Fusion Frame Work

Image fusion is the process, which gathers significant details from two or more images. In the image fusion important information is selected from multiple images. Implementation of fusion of images is carried out either in spatial or in transforms domains. In this work, fusion is done in both domains to get better performance. Energy of decomposed bands of NSCT is used to select important bands in NSCT based image fusion. Energy of decomposed bands of DWT is used to select important bands in DWT based image fusion. Fused images of NSCT and DWT are further fused by using spatial domain technique. In spatial fusion ESOP values are taken into consideration to perform fusion. Experiments are done on several medical images .Experiments show that, the proposed method is giving perceptually meaningful fused images. Image metrics like entropy, edge based similarity measure and quality of mutual information have been used for the assessment of performance of the work. In this research work, two medical images (CT, MRI), after pre-processing, will be merged according to the wavelet and NSCT transformations using energy fusion techniques to generate two independent fusion images that will be merged again using spatial domain to get the desired output. In this way the large amount of comprehensive information can be presented in the merged image, all the comprehensive information obtained from the two medical images appears in the final output. The experimental outcomes on different CT and MRI images are analyzed qualitatively and quantitatively. Image fusion has been implemented in the various applications like remote sensing, space research, defence, medical imaging etc. The performance parameters show remarkable improvements.

Effect of Geometric Sharpness on Translucent Material Perception

When judging optical properties of a translucent object, humans often look at sharp geometric features such as edges and thin parts. Analysis of the physics of light transport shows that these sharp geometries are necessary for scientific imaging systems to be able to accurately measure the underlying material optical properties. In this paper, we examine whether human perception of translucency is likewise affected by the presence of sharp geometry, by confounding our perceptual inferences about an object’s optical properties. We employ physically accurate simulations to create visual stimuli of translucent materials with varying shapes and optical properties under different illuminations. We then use these stimuli in psychophysical experiments, where human observers are asked to match an image of a target object by adjusting the material parameters of a match object with different geometric sharpness, lighting geometry, and 3D geometry. We find that the level of geometric sharpness significantly affects perceived translucency by the observers. These findings generalize across a few illuminations and object shapes. Our results suggest that the perceived translucency of an object depends on both the underlying material optical parameters and 3D shape. We also conduct analyses using computational metrics including (luminance-normalized) L2, structural similarity index (SSIM), and Michelson contrast. We find that these image metrics cannot predict perceptual results, suggesting low level image cues are not sufficient to explain our results.