Multilevel algorithm for color clustering of images

The fast multilevel algorithm to cluster color images (MACC – Multilevel Algorithm for Color Clustering) is presented. Currently, several well-known algorithms of image clustering, including the k‑means algorithm (which is one of the most commonly used in data mining) and its fuzzy versions, watershed, region growing ones, as well as a number of new more complex neural network and other algorithms are actively used for image processing. However, they cannot be applied for clustering large color images in real time. Fast clustering is required, for example, to process frames of video streams shot by various video cameras or when working with large image databases. The developed algorithm MACC allows the clustering of large images, for example, FullHD size, on a personal computer with an average deviation from the original color values of about five units in less than 20 milliseconds, while a parallel version of the classical k‑means algorithm performs the clustering of the same images with an average error of more than 12 units for a time exceeding 2 seconds. The proposed algorithm of multilevel color clustering of images is quite simple to implement. It has been extensively tested on a large number of color images.

Recognition of the layout of colored yarns in yarn-dyed fabrics

The layout of colored yarns in yarn-dyed fabrics is a significant part of designing and production in the textile industry, which is still analyzed manually at present. Existing methods based on image processing have some limitations in accuracy and stability. Therefore, an automatic method is proposed to recognize the layout of colored yarns and some other basic fabric structure parameters: the fabric density and weave pattern. First, a large dataset with fabric structure parameters is constructed. The fabric images are captured by a wireless portable device. Then the yarns and floats are accurately located using a novel multi-task and multi-scale convolutional neural network. Finally, a density-based color clustering algorithm is proposed to recognize the layout of colored yarns. The results of extensive experiments show that the proposed method can automatically identify the basic structure parameters with high effectiveness and robustness.

“TNOs are Cool”: A survey of the trans-Neptunian region

The goal of this work is to determine the physical characteristics of resonant, detached and scattered disk objects in the trans-Neptunian region, observed mainly in the framework of the “TNOs are Cool” Herschel open time key programme. Based on thermal emission measurements with the Herschel/PACS and Spitzer/MIPS instruments, we determine size, albedo, and surface thermal properties for 23 objects using radiometric modeling techniques. This is the first analysis in which the physical properties of objects in the outer resonances are determined for a notable sample. In addition to the results for individual objects, we compared these characteristics with the bulk properties of other populations of the trans-Neptunian region. The newly analyzed objects show a large variety of beaming factors, indicating a diversity of thermal properties, and in general they follow the albedo-color clustering identified earlier for Kuiper belt objects and Centaurs, further strengthening the evidence for a compositional discontinuity in the young Solar System.

Research on Digital Camouflage Pattern Generation Algorithm Based on Adversarial Autoencoder Network

In the past, most of the digital camouflage used textural features to extract the configuration features of spots in gray images, unable to effectively utilize the position relationship between color information. In order to overcome this shortcoming, a new digital camouflage pattern design model was proposed based on the model of adversarial autoencoder network. Firstly, the complexity and performance of several main color extraction algorithms were analyzed and compared, and combined with AFK-MC2 algorithm and color similarity coefficient, a fast camouflage main color clustering method was proposed. Then a deep convolution adversarial autoencoder network was designed to extract and describe the configuration features of the spots in background pattern. In order to diffuse pixel spot and achieve the effect of spatial color blending, a morphological processing algorithm was proposed to process the generated camouflage patterns. Finally, two sets of grassland and woodland datasets were established, respectively. The influence of the number of latent variables of network on the training process was tested on the dataset, and the number of camouflage feature descriptions was determined to be greater than or equal to 10. In order to verify the effectiveness of the generated camouflage, the spots in background region and target region were randomly selected, and the Euclidean distance between the feature parameters of these spots was calculated. Both the visual and experimental results demonstrate that the generated spots have high fusion with the background.

An Improved Seed Point Selection-Based Unsupervised Color Clustering for Content-Based Image Retrieval Application

The images involved in the content-based image retrieval (CBIR) applications are collectively represented by features such as color, texture and shape. The precision of the CBIR application relies on the key features used in image representation and its similarity measure. In CBIR, dominant color feature extraction is affected by the predefined intervals used in color quantization. The proposed work mainly concentrates on extracting the dominant color information of the image using the clustering process. The clustering process is initiated by the proposed seed point’s selection approach. This approach derives the number of seed points using the first order statistical measure and maximum range of the distributed pixel values. Moreover, this work gives equal priority to dominant color and its occurrence information in calculating the similarity between query and database images. Finally, the standard databases such as SIMPLIcity, Corel-10k, OT-scene, Oxford flower and GHIM are taken to investigate the performance of the proposed dominant color based image retrieval application.