Contour Learning and Diffusive Processes for Colour Perception

The human brain processes enormous volumes of high-dimensional data for everyday perception. To humans, a picture is worth a thousand words, but to a machine, it is just a seemingly random array of numbers. Although machines are very fast and efficient, they are vastly inferior to humans for everyday information processing. Algorithms that mimic the way the human brain computes and learns may be the solution. We present a theoretical model based on the observation that images of similar visual perceptions reside in a complex manifold in a low-dimensional image space. The perceived features are often highly structured and hidden in a complex set of relationships or high-dimensional abstractions. To model the pattern manifold, we present a novel learning algorithm using a recurrent neural architecture. The brain memorizes information using a dynamical system made of interconnected neurons. Retrieval of information is accomplished in an associative sense. It starts from an arbitrary state that might be an encoded representation of a visual image and converges to another state that is stable. The stable state is what the brain remembers. In designing a recurrent neural architecture, it is usually of prime importance to guarantee the convergence in the dynamics of the network. We propose to modify this picture: if the brain remembers by converging to the state representing familiar patterns, it should also diverge from such states when presented with an unknown encoded representation of a visual image belonging to a different category. That is, the identification of an instability mode is an indication that a presented pattern is far away from any stored pattern and therefore cannot be associated with current memories. These properties can be used to circumvent the plasticity-stability dilemma by using the fluctuating mode as an indicator to create new states. We capture this behavior using a novel neural architecture and learning algorithm, in which the system performs self-organization utilizing a stability mode and an instability mode for the dynamical system. Based on this observation we developed a self-organizing line attractor, which is capable of generating new lines in the feature space to learn unrecognized patterns. Experiments performed on various databases show that the proposed nonlinear line attractor is able to successfully associate patterns and it provides better association when compared to other state of the art techniques. It shows that the proposed model is able to create nonlinear manifolds in a multidimensional feature space to distinguish complex patterns.

Memory degradation induced by attention in recurrent neural architectures

Image retrieval is the process of finding images in the database that are similar to the query image by measuring how close the feature values of the query image are to other images. Image retrieval is currently dominated by approaches that combine several different representations or features. The optimal weight of each feature is needed in combining the image features such as color features, texture features, and shape features. In this study, we use a multi-layer perceptron artificial neural network (MLP) method to obtain feature weights automatically and simultaneously look for optimal weights. The color moment is used to find nine color features, Gray Level Co-occurrence Matrix (GLCM) to find four texture features, and Hu Moment to find seven shape features totaling 20 neurons and all of these features become the input layer in our MLP network. Three neurons in output layers become the automatic weight of each feature. These weights are used to combine each feature's role in obtaining the relevant image. Euclidean Distance is used to measure similarity. The average precision values obtained using automatic feature weights are 93.94% for the synthetic dataset, 91.19% for the Coil-100 dataset, and 54.31% for the Wang dataset. These results have an average difference of 5.06% with the target so automatic feature weighting works well. This value is obtained at a hidden layer size of 11 and a learning rate of 0.1. In addition, the use of automatic feature weighting gives more accurate results compared to manual feature weighting.

Radiomics has shown potential in disease diagnosis, but its feasibility for non-small cell lung carcinoma (NSCLC) subtype classification is unclear. This study aims to explore the diagnosis value of texture and colour features from positron emission tomography computed tomography (PET-CT) images in differentiation of NSCLC subtypes: adenocarcinoma (ADC) and squamous cell carcinoma (SqCC). Two patient cohorts were retrospectively collected into a dataset of 341 18F-labeled 2-deoxy-2fluoro-d-glucose ([18F] FDG) PET-CT images of NSCLC tumours (125 ADC, 174 SqCC, and 42 cases with unknown subtype). Quantification of texture and colour features was performed using freehand regions of interest. The relation between extracted features and commonly used parameters such as age, gender, tumour size, and standard uptake value (SUVmax) was explored. To classify NSCLC subtypes, support vector machine algorithm was applied on these features and the classification performance was evaluated by receiver operating characteristic curve analysis. There was a significant difference between ADC and SqCC subtypes in texture and colour features (P < 0.05); this showed that imaging features were significantly correlated to both SUVmax and tumour diameter (P < 0.05). When evaluating classification performance, features combining texture and colour showed an AUC of 0.89 (95% CI, 0.78–1.00), colour features showed an AUC of 0.85 (95% CI, 0.71–0.99), and texture features showed an AUC of 0.68 (95% CI, 0.48–0.88). DeLong’s test showed that AUC was higher for features combining texture and colour than that for texture features only (P = 0.010), but not significantly different from that for colour features only (P = 0.328). HSV colour features showed a similar performance to RGB colour features (P = 0.473). The colour features are promising in the refinement of NSCLC subtype differentiation, and features combining texture and colour of PET-CT images could result in better classification performance.

We address the problem of localizing objects using color, texture and shape. Given a handrawn sketch for querying an object shape, and its color and texture, the algorithm automatically searches the database images for objects which meet the query attributes. The database images do not need to be presegmented or annotated. The proposed algorithm operates in two stages. In the first stage, we use local texture and color features to find a small number of candidate images, and identify regions in the candidate images which share similar texture and color as the query example. To speed up the processing, the texture and color features are directly extracted from the Discrete Cosine Transform (DCT) compressed domain. In the second stage, we use a deformable template matching method to match the query shape to the image edges at the locations which possess the desired texture and color attributes. This algorithm is different from the other content-based image retrieval algorithms in that: (i) no presegmentation of the database images is needed, and (ii) the color and texture features are directly extracted from the compressed images. Experimental results show that substantial computational savings can be achieved utilizing multiple image cues.

Object localization using color, texture and shape

In this paper, a new unsupervised hierarchical approach to textured color images segmentation is proposed. To this end, we have designed a two-step procedure based on a grey-scale Markovian over-segmentation step, followed by a Markovian graph-based clustering algorithm, using a decreasing merging threshold schedule, which aims at progressively merging neighboring regions with similar textural features. This hierarchical segmentation method, using two levels of representation, has been successfully applied on the Berkeley Segmentation Dataset and Benchmark (BSDB, Martin et al., 2001). The experiments reported in this paper demonstrate that the proposed method is efficient in terms of visual evaluation and quantitative performance measures and performs well compared to the best existing state-of-the-art segmentation methods recently proposed in the literature.

In this chapter, a content-based image retrieval technique based on the concept of region-based image retrieval has been described. This technique integrates color, texture, and shape features using local binary patterns (LBPs). In this technique, the image is divided into a fixed number of blocks and from each block LBP-based color, texture, and shape features are computed. The color and texture features are extracted using LBP histograms of quantized color image and gray-level images, respectively. Shape features are computed using the binary edge map obtained using Sobel edge detector from each block. All three features are combined to make a single completed binary region descriptor (CBRD) represented in the LBP way. To support region-based retrieval, an effective region code-based scheme is employed. In this technique, the spatial relative locations of objects are also considered to increase the retrieval accuracy.

Content-Based Image Retrieval (CBIR) allows to automatically extracting target images according to objective visual contents of the image itself. Representation of visual features and similarity match are important issues in CBIR. Color, texture and shape information have been the primitive image descriptors in content-based image retrieval systems. This paper presents a fast and efficient image indexing and search system based on color and texture features. The color features are represented by combines 2-D histogram and statistical moments and texture features are represented by combines the direction measure and a gray level co-occurrence matrix (GLCM). The detailed experimental analysis is carried out using precision and recall ondataset Wang. The time analysis is also performed to compare processing speeds of the proposed method with the existing similar best. The experimental results demonstrated that proposed method achieved a better image recognition, and the accuracy of classification based on this method has been significantly improved. The performance is measured in terms of recall and precision; also the obtained performances are compared with several state-of-the-art algorithms and showed that our algorithm is simple, fast, and efficient in terms of results and memory. Keywords: Co-occurrence Matrix, Direction Measure, 2-D Histogram, Statistical Moments, CBIR, GLCM

Inspection of paddy seed varietal purity using machine vision and multivariate analysis

With the emerge of multimedia, communication and processing, a typical image database has a large volume of data. It is an essential to build an efficient retrieval system to browse through the entire database. In this paper we proposed precise Relevance Feed Back (RFB) Content Based Image Retrieval (CBIR) using multiple features based on interactive retrieval approach which will extensively reduces the semantic gap between low-level features and high-level semantics. Relevance Feed Back improves the retrieval accuracy of Content Based Image Retrieval(CBIR) by modifying the query based on the user's feedback in which the user can select the most relevant images and provide a weight of preference for each relevant image. In our approach we extracted multiple features by integrating color, texture and shape features with Relevance Feed Back. We extracted color features by cumulative histogram, texture feature are exploited by using Color co-occurrence matrix (CCM) and we describe the shape information contained in an image on the basis of its significant edges. We used Euclidean distance as the similarity measure. In our experiment we used Corel real-world image databases with 1000 images, divided into 10 categories, each category containing 100 images including landscapes, animals, plants, monuments, transport (cars, planes). Our results reflects that integrating color, texture and shape for retrieving similar images improves the retrieval performance with respect to retrieval accuracy and recall rate as compared with the classical color feature. During the experiment we also observed that incorporating Relevance Feed Back (RFB) with multiple features outperforms in comparison with multifeature vector without Relevance Feed Back method.

Rapid development of multimedia technologies made Content Based Image Retrieval (CBIR) an energetic research area for the researchers of multimedia domain. Texture and color features have been the primal descriptors for images in the field of CBIR. This paper proposed a new CBIR system by combining the both color and texture features. Color Layout Descriptor (CLD) from MPEG-7 is used for the color feature extraction while, Mean, variance, skewness, Kurtosis, energy and entropy are used as texture descriptors. Experiments are performed on Coral Database. The results of the proposed method namely CLD-fos are compared with the four well reputed systems (i.e. SIMPLIcity, Histogram based, FIRM, and Variance Segment etc) from the industry. The results of the CLD-fos demonstrated high accuracy rate than the previous systems during the simulations. The proposed CLD-fos achieved significant performance in terms of accuracy.

Previous studies with the title of the login authentication e-library with method of CBIR for matching face, have proved reaching the level of accuracy about 75%. This multiple verification of QR-code/QR-CMBS this process data among other things the identity ID, fingerprint patterns and pattern signatures. Each user can have a QR-CMBS, which is used to login to the e-library. This research-oriented system development with application authentication login with QR code/QR-QR, Data of the CMBS will store data from bineri identity ID, fingerprint patterns and pattern signatures.The advantage of Retrieval CBIR is the popularity and test result with a high degree of accuracy and time parameters. The results obtained from QR-CMBS every training, i.e. classify and determine the value of fingerprint patterns and signatures for each label. Feature extraction results are temporarily stored in the session database and compare the features that are stored in the database image classification. The most similar classification results will be displayed, i.e. QR-CMBS, fingerprints and signatures, as well as verification of login. The application login authentication system of e-library uses to calculate the similarity of this research, will be able to extract the feature of colour, texture and edge of a multiple verification of QR-code/ QR-CMBS, fingerprint and signature by using the Prewitt gradient. The result of the extraction process feature is then used by the software in the learning process and calculates the similarity. Learning image contained in 3 classes features a picture that is stored in the database query 100 png images and the image of the sample test with the size 400 x 400. The results showed that the combination of the Prewitt filter extraction gradient magnitude. Verification data classification compared to the three classes, namely QR-CMBS, fingerprints and signatures contained in the database. Response time to find the most CMBS-QR is similar to 10 sample data, giving the effect of a higher degree of accuracy that is 97%.

Terrorist events have a substantial emotional impact on the population, and understanding these effects is very important to design effective assistance programs. However, investigating community-wide traumas is a complex and costly task, where most challenges are related to the data collection process. Social media has been used as a relevant source of data to investigate people’s sentiments and ideas. In this article, we study the emotional reactions of Twitter users regarding two terrorist events that occurred in the United Kingdom. The contributions are twofold: a) we experiment two deep learning architectures to develop an emotion classifier, and b) we develop an analysis on tweets related to terrorist events to underst and whether there is an emotional shift due to a terrorist attack andwhether the emotional reactions are dependent on the event, or on the demographics of the users. Both models, based on convolutional and recurrent neural architectures, presented very similar performances. The analyses revealed an emotion shift due to the events and a difference in the reactions to each specific event, where gender is the most significant factor.

This research proposed the identification of chicken freshness level based on its color and texture features. Color Features used are the RGB (Red, Green, and Blue) and HSV (Hue, Saturation, Value) channel histogram value. texture features used are GLCM (Grey Level Co-Occurrence Matrix), Gabor kernel, and HOG (Histogram of Oriented Gradients). The freshness level of a chicken meat is categorized into three labels, fresh (0-4 hours after slaughtered), medium-fresh (4-6 hours after slaughtered), and not-fresh (more than 6 hours after slaughtered). The experiments will identify the freshness using several classification methods and different camera resolution and magnification. The highest classification accuracy using SVM (Support Vector Machines) achieves 58,33% with a smartphone camera, 98% with a webcam camera, and 79.1% with a 200 magnification digital microscope. From the experiment results, we can conclude that using webcam camera with normal resolution have better classification accuracy compared with a 200 magnification digital microscope or standard smartphone camera. It is also shown that SVM is superior compared with other methods tested in this experiments which are Decision Tree and Naive Bayes.