CIE 1931 Research Articles

A contextualized approach for segmentation of foliage in different crop species

Computer vision algorithms represent an important tool in agriculture. Particularly, foliage segmentation is a fundamental step to achieve accurate estimations in more complex visual analysis such as plant health monitoring, weed detection, growth estimation, and flower and fruit classification. Unfortunately, this complexity leads to the problem of generalizing between datasets, an issue that has not yet been addressed by the state of the art. Our work aims at filling this gap by investigating the role of vegetative indexes and color spaces in different formulations of machine learning algorithms. To this end, we have considered datasets observing variations with respect to crop species, leaf color, acquisition settings and illumination. Furthermore, we performed a comparison with different state of the art approaches that include both thresholding and machine learning. From our analysis, we have proposed a new formulation that consists of combining the CIE Luv color space and support vector machines in order to benefit from contextualized information obtained through neighboring pixels. Experiments show that our approach achieves the best results.

Various Color Models and Color Feature Indexes to Predict Fruit Ripeness Stages

In general, farmers who grow fruits such as strawberry or tomato in greenhouses are very eager to predict the optimal harvesting time for fruits in advance. Therefore, in order to solve these farmers needs, this paper suggests the appropriate color indexes representing a ripeness of fruit images which are essential to develop a system that can automatically determine the degree of maturity of fruit. First, we consider the definitions and properties of three color models such as CIE XYZ, CIE LAB, and CIE LUV that are commonly used to represent fruit images. Second, we examine the color components and the extracting method of color feature vectors that is most suitable to accurately determine the ripening stage of the fruit. From the experimental results, the color components that can accurately distinguish the stages of ripening of strawberry were Y component in XYZ color model, A component in LAB color model, and V component in LUV color model. Also, it can be seen that the average, skewness and kurtosis of color values can be used as color feature vectors effective for determining the degree of ripening of strawberry. Therefore, the results of our study are expected to be useful to develop a system that can automatically determine the harvesting time of fruits.

An effective color texture image segmentation algorithm based on hermite transform

In this paper, an efficient color texture image segmentation approach is proposed. The proposed approach uses color and texture information independently. The color information is obtained by converting the RGB color space to Luv color space and each color component is considered as a color descriptor. For texture descriptors, Hermite transform is considered. Hermite transform uses the Hermite filters which are formed by the product of Hermite polynomials with Gaussian function. Instead of using all Hermite filters, a filter selection process is adopted to obtain optimal filters. A feature image is constructed based on the magnitude of each filter response. A region smoothing procedure is employed for both the color components and the feature image in order to make the region smoother while preserving the edge information. To this end, weighted least square edge-preserving filtering is used. Comprehensive experiments were conducted to demonstrate the efficiency of the proposed method, using the Berkeley segmentation dataset.

Lip segmentation using automatic selected initial contours based on localized active contour model

With the rapid development of artificial intelligence and the increasing popularity of smart devices, human-computer interaction technology has become a multimedia and multimode technology from being computer-focused to people-centered. Among all ways of human-computer interactions, using language to interact with machines is the most convenient and efficient one. However, the performance of audio speech recognition systems is not satisfied in a noisy environment. Thus, more and more researchers focus their works on visual lip reading technology. By extracting lip movement features of speakers rather than audio features, visual lip reading systems can get superior results when noises and interferences exist. Lip segmentation plays an important role in a visual lip reading system, since the segmentation result is crucial to the final recognition accuracy. In this paper, we propose a localized active contour model-based method using two initial contours in a combined color space. We apply illumination equalization to original RGB images to decrease the interference of uneven illumination. A combined color space consists of the U component in CIE-LUV color space and the sum of C2 and C3 components of the image after discrete Hartley transform. We select a rhombus as the initial contour of a closed mouth, because it has a similar shape to a closed lip. For an open mouth, we utilize a combined semi-ellipse as the initial contours of both outer and inner lip boundaries. After attaining the results of each color component separately, we merge them together to obtain the final segmentation result. From the experiment, we can conclude that this method can get better segmentation results compared with the method using a circle as the initial contour to segment gray images and images in combined color space, especially for open mouth. An extremely obvious advantage of this method is the results of open mouth excluding internal information of mouth such as teeth, black holes, and tongue, because of the introduction of the inner initial contour.

A hybrid colour model based land cover classification using random forest and support vector machine classifiers

Land cover monitoring using remotely sensed data requires robust classification methods for the accurate mapping of complex land cover and land use categories. Classification is a supervised learning method which maps a data item into predefined classes. Colour is an important feature used in image classification since humans tend to distinguish images mostly based on colour feature. This paper proposes a hybrid colour model for land cover classification in which colour features are extracted by combining the hue (H) values of HSV colour space and luminance (L) values of LUV colour space. The extracted features are trained and tested with random forest (RF) and support vector machine (SVM) classifiers. The performance of the proposed hybrid colour model is compared with the existing HSV colour space model using RF and SVM classifiers based on several metrics such as accuracy, sensitivity, specificity and f-score. Hyper spectral dataset of Pavia University and an IRS LISS IV orthorectified dataset are chosen as the input image for this experiment.

A hybrid colour model based land cover classification using random forest and support vector machine classifiers

Land cover monitoring using remotely sensed data requires robust classification methods for the accurate mapping of complex land cover and land use categories. Classification is a supervised learning method which maps a data item into predefined classes. Colour is an important feature used in image classification since humans tend to distinguish images mostly based on colour feature. This paper proposes a hybrid colour model for land cover classification in which colour features are extracted by combining the hue (H) values of HSV colour space and luminance (L) values of LUV colour space. The extracted features are trained and tested with random forest (RF) and support vector machine (SVM) classifiers. The performance of the proposed hybrid colour model is compared with the existing HSV colour space model using RF and SVM classifiers based on several metrics such as accuracy, sensitivity, specificity and f-score. Hyper spectral dataset of Pavia University and an IRS LISS IV orthorectified dataset are chosen as the input image for this experiment.

Detection and segmentation of retinal blood vessel in digital RGB and CIELUV color space fundus images

The identification of retinal blood vessels is very important but crucial task to analyze the severity of the retinal diseases such as diabetic retinopathy, macular degeneration, central retinal vein occlusion, central retinal artery occlusion, retinal detachment and branch retinal vein occlusion. It is evident that huge number of computer based automated algorithms are developed for the accurate detection of blood vessels and optical disc. Most of the work utilizes the retinal fundus images in RGB color space. The proposed work implements the detection and segmentation of retinal blood vessel in RGB and device independent CIELUV color space. The proposed work for the segmentation retinal blood vessel is based on adaptive histogram equalization, median filtering and morphological operations.

Combination of Brightness Preserving Bi-Histogram Equalization and Discrete Wavelet Transform using LUV Color Space for Image Enhancement

Combination of Brightness Preserving Bi-Histogram Equalization and Discrete Wavelet Transform using LUV Color Space for Image Enhancement

Local Phase Features in Chromatic Domain for Human Detection

In this paper, a new descriptor based on phase congruency concept and LUV color space features is presented. Since the phase of the signal conveys more information regarding signal structure than the magnitude and the indispensable quality of the color in describing the world around us, the proposed descriptor can precisely identify and localize image features over the gradient based techniques, especially in the regions affected by illumination changes. The proposed features can be formed by extracting the phase congruency information for each pixel in the three-color image channels. The maximum phase congruency values are selected from the corresponding color channels. Histograms of the phase congruency values of the local regions in the image are computed with respect to its orientation. These histograms are concatenated to construct the proposed descriptor. Results of the experiments performed on the proposed descriptor show that it has better detection performance and lower error rates than a set of the state of the art feature extraction methodologies.

Scheme for unsupervised colour–texture image segmentation using neutrosophic set and non‐subsampled contourlet transform

The process of partitioning an image into some different meaningful regions with the homogeneous characteristics is called the image segmentation which is a crucial task in image analysis. This study presents an efficient scheme for unsupervised colour–texture image segmentation using neutrosophic set (NS) and non-subsampled contourlet transform (NSCT). First, the image colour and texture information are extracted via CIE Luv colour space model and NSCT, respectively. Then, the extracted colour and texture information are transformed into the NS domain efficiently by the authors’ proposed approach. In the NS-based image segmentation, the indeterminacy assessment of the images in the NS domain is notified by the entropy concept. The lower quantity of indeterminacy in the NS domain, the higher confidence and easier segmentation could be achieved. Therefore, to achieve a better segmentation result, an appropriate indeterminacy reduction operation is proposed. Finally, the K-means clustering algorithm is applied to perform the image segmentation in which the cluster number K is determined by the cluster validity analysis. To show the effectiveness of their proposed method, its performance is compared with that of the state-of-the-art methods. The experimental results reveal that their segmentation scheme outperforms the other methods for the Berkeley dataset.

RELATIONSHIP BETWEEN SOIL MOISTURE CONTENT IN PADDY FIELD AND ITS IMAGE TEXTURE

The aim of this study is to identify the relationship between soil moisture content and its image texture. Soil image was captured and converted into CIELUV color space. These images were later used to develop two dimensional gray level co-occurrence matrix. Eight texture features extracted from gray level co-occurrence matrix namely mean, variance, homogeneity, dissimilarity, entropy, contrast, second moment and correlation was used for the analysis. The results has shown that the image texture properties can be used to relate with soil moisture content, where variance, homogeneity, dissimilarity, entropy, contrast, second moment and correlation gave significant responds to the moisture content. The highest value of correlation was gathered from entropy with r = -0.522.

Multifocus color image sequence fusion based on mean shift segmentation.

This paper presents a region-based technique for fusion of a multifocus color image sequence in the LUV color space. First, mean shift segmentation was applied on the weighted average image of the image sequence to obtain the fusion reference areas. Second, for each segmented area, the well-known modified Laplacian (LAP2) was used as a focus measure to select the clearest parts within the image sequence and then a final image focused with all parts can be generated. Mutual information, QAB/F metric, entropy, standard deviation, image sharpness metric, image contrast metric, average gradient, and spatial frequency were adopted to assess the quality of the fused image. Experiments carried out using standard image sequences from HeliconSoft demonstrated that the results obtained through our technique offer good performance. The proposed technique can be used to extend the depth of field (DOF) of a camera system effectively.

Segmentation of Optic Disc and Optic Cup to Calculate CDR using Kekre’s LUV Color Space for Detecting Glaucoma

Glaucoma is the term applied to an eye disease that gradually result in loss of vision by permanently harming the optic nerve, the nerve that transmits visual pictures to the brain. Glaucoma is the second leading cause of blindness worldwide. The retina is the deepest layer in the eye and the retinal nerve filaments transmit the visual sign from the photoreceptors in the eye to the mind through the pack going out of the eye, known as the optic nerve. Glaucoma leads to consistent and expedient harm of the retinal nerve fiber layer and consequently can prompt lasting visual impairment. Thus the finding of glaucoma at a prior stage is imperative for its treatment. Current tests using intraocular pressure (IOP) are not sufficiently delicate for populace based glaucoma screening. Glaucoma can be detected by finding out CDR. CDR Gives Cup to Disc ratio of Optic Cup and Optic Disc. CDR will be finding out by calculating Vertical diameter of Optic cup (OC) and Vertical diameter of Optic disc (OD). To find out Optic Disc and Optic Cup, Kekre’s LUV color space will be applied along with k-means clustering. On the off chance that the CDR proportion surpasses threshold level 0.6 it shows high risk of Glaucoma for the tried patient. General Terms CDR, VDD, VCD

Measuring saturation

For most color spaces, there is a measure for the saturation of a color. Most of these measures are not ordinally equivalent and they are mainly based on color discrimination experiments. It is unclear how well they correspond to human supra-threshold perception of saturation. We chose ten color comparison directions and three standard colors at a fixed luminance of 30 cd/m2. In each trial, two color patches were shown for 750 ms on a color calibrated monitor. One patch always had the color of one of the three standard colors while the color of the other patch was sampled by means of an adaptive algorithm from one of the ten comparison directions. Participants were asked to decide which of the two colors was more saturated by pressing one of two buttons. Background luminance was 10 cd/m2. We computed the point of subjective equality (PSE) for each of the standards in all of the ten color directions. We compared these PSEs to those predicted by different saturation measures defined in the CIECAM02, HSV, DKL, LAB, LUV, and CIE 1931 xyY color spaces. The measures differ in their ability to predict human perception of saturation, with LUV and LAB being closest to the observers' judgments and just above the best level allowed for by the inter-observer consistency. The measures defined in CIE xyY and HSV performed worst. The various measures of saturation differ in how well they capture human perception of saturation. For a constant level of luminance, human perception of saturation can best be described by the measures defined in LUV and LAB color space. Meeting abstract presented at VSS 2015

Automatic building detection based on Purposive FastICA (PFICA) algorithm using monocular high resolution Google Earth images

This paper proposes an improved FastICA model named as Purposive FastICA (PFICA) with initializing by a simple color space transformation and a novel masking approach to automatically detect buildings from high resolution Google Earth imagery. ICA and FastICA algorithms are defined as Blind Source Separation (BSS) techniques for unmixing source signals using the reference data sets. In order to overcome the limitations of the ICA and FastICA algorithms and make them purposeful, we developed a novel method involving three main steps: 1-Improving the FastICA algorithm using Moore–Penrose pseudo inverse matrix model, 2-Automated seeding of the PFICA algorithm based on LUV color space and proposed simple rules to split image into three regions; shadow+vegetation, baresoil+roads and buildings, respectively, 3-Masking out the final building detection results from PFICA outputs utilizing the K-means clustering algorithm with two number of clusters and conducting simple morphological operations to remove noises. Evaluation of the results illustrates that buildings detected from dense and suburban districts with divers characteristics and color combinations using our proposed method have 88.6% and 85.5% overall pixel-based and object-based precision performances, respectively.

필름 색 오버레이의 광학적 특성과 사용자 선택을 반영한 컴퓨터 구현성 분석

광과민성 증후군자를 위한 가상 색 오버레이의 구현성을 검토하고 요구사항을 도출하였다. 대표적 필름 오버레이 2 종류의 광학적 특성을 측정하고, 그 특성을 구현하여 컴퓨터 및 모바일 IT기기에 구현할 <TEX>${\alpha}$</TEX>값과 원 색도를 도출하였다. 주변조명을 암실환경과 명실환경으로 구분하여 주변조명이 가상 색 오버레이 색도에 미치는 영향을 분석하였다. 하나의 원 색도(CIE Luv)를 만족하는 <TEX>${\alpha}$</TEX>와 RGB값의 조합이 다수 존재하는 것을 발견하였고, 사용자는 눈의 편안함과 선명도를 기준으로 한 조합을 선택할 수 있다. 주변조명은 100 lx차이에서 디스플레이 장치의 색도변화는 거의 없으나, 밝기 차이가 평균 <TEX>$37.8395cd/m^2$</TEX>(표준편차 2.0059) 발생하였다. The feasibility of a virtual colored overlay is based on constraints of computer and user preference and it is evaluated for people with Scotopic Sensitivity Syndrome. We draw <TEX>${\alpha}$</TEX> and chromaticities of virtual colored overlay which will be implemented on IT devices by measuring optical characteristics of 2 representative film overlays and simulating them. We analyzed the influence of two illuminating conditions on the virtual colored overlay. We find that it is possible to get the multiple combinations of alpha and source RGB values to match a chromaticity which is presented in CIE-Luv color space under ideal white condition of display, a user can select one among them with respect to clarity and comfort. Under 100 lx difference of illumination conditions, the changes of chromaticities are negligible, but luminances are increased <TEX>$37.8cd/m^2$</TEX> average(std 2.006) at high illumination condition.

A Novel Edge Detection Algorithm of Color Image

This paper proposes a novel algorithm of edge detection using LUV color space. Firstly, according to peer group filtering (PGF), a nonlinear algorithm for image smoothing and impulse noise removal in color image is used. Secondly, color image edges in an image are obtained automatically by combining an improved isotropic edge detector and a fast entropy threshold technique. Thirdly, according to color distance between the pixel and its eight neighbor-pixels, color image edges can further be detected. Finally, the experiment demonstrates the outcome of proposed algorithm in color image edge detection.

Selective Search Method for Object Localization and Detection using Wavelets and Hierarchical Segmentations

This article proposes a selective search method for object localization in natural images by applying image multi-segmentation, image scaling, and heuristics. The method increases the number of generated windows that delimitate the area of an object with an accuracy superior to 50%. Over-segmentation is applied on original size images in order to locate small objects, and it is also applied over scaled images because these can still be over-segmented. This process produces less regions on areas with many textures. The over-segmentation was applied using the CIE Luv color model, and using the H and the I channels of the HSI model. The proposed method is category independent and allows the location of objects with heterogeneous characteristics by using heuristics and hierarchical segmentation. The proposed method produces 9, 366 windows per image covering 96.78% of the objects in the PASCAL VOC 2007 test image collection, increasing in 0.8% the localization results reported in the state of the art.

Image Compression using Cosine Slant Hybrid Wavelet Transform with Assorted Color Spaces

Gigantic size of multimedia data and images being generated stored and transmitted across the digital web today, which has raised new research concerns in computing field. One of such concerns is compression of such data with minimal quality distortion. In recent work, the hybrid wavelet transforms (HWT) are proven to be better in image compression as compared to the constituent transforms considered individually [1, 2]. Then further the proportion wise performance comparison of the constituent transforms considered for generation of HWT is also studied [3, 4]. Here the paper presents the performance appraise of various color spaces along with the conventional RGB color space in image compression using cosine-slant HWT with varying proportions of constituent cosine and slant transforms. The experimentation is done on a test-bed having 15 images of various sizes and varied compression ratios. The results show that LUV color space with 1:4 proportion of cosine-slant transform in HWT gives better quality of compression for higher compression ratios (90% and 95%). For 85% compression ratio, LUV color space with 1:1 proportion of cosine-slant transforms in HWT gives better compression quality. While in lower compression ratios (60% to 80%), the RGB color space gives better compression with 1:1 proportion of cosine-slant transforms in HWT. General Terms Image Compression.

Color Image Quantization for Reduction of Quantization Contours and Preservation of Perceptually Sensitive Colors

The conventional methods for color image quantization are aimed at obtaining the resulting image with a minimum mean squared error (MSE). Yet human perceptual satisfaction is not always related to the MSE. This article presents a color quantization method that uses a lightness enhancement factor for better representation of the gradation and a perceptual threshold for preservation of perceptually sensitive colors in the CIELUV color space. The experimental results show that the proposed method provides better visual quality when representing the gradation and better preserves the perceptually sensitive colors compared to the conventional methods even if the MSE of the proposed method is slightly higher than those of the conventional methods.