Skin Color Regions Research Articles

Preferred skin reproduction of different skin groups

To accurately reproduce preferred skin colour is an important goal in digital image colour reproduction. A psychophysical experiment was conducted to specify the preferred skin colour for different skin types. Ten original facial images were captured to cover different skin types, including Caucasian, Chinese, South Asian, African, different genders and different ages. For each original image, 49 rendered images, uniformly sampled within the skin colour ellipsoid in CIELAB colour space, were used to morph the skin colours. Thirty observers from each of three ethnic backgrounds, Caucasian, Chinese and South Asian, participated in the experiment to investigate ethnic differences. Ellipsoid models were developed to specify preferred skin colour regions and centres for each original image. These results can be used to improve skin colour reproduction of colour imaging products, for example on mobile phones, for different skin types.

Study on face detection based on an improved Gaussian skin color model

An improved Gaussian skin color model is proposed for face detection applications. It overcomes the issue that the traditional models become ineffective in extracting the skin color region in the presence of diverse skin colors and illuminations. It is a two-dimensional gaussian mixture model based on H-SV (i.e., an improved HSV) and YCbCr color spaces. Experimental results from several face image databases show that the detection accuracy is improved by 35.8% and 6.3%, respectively, compared to a traditional gaussian skin model and Adaboost algorithm.

Multiskin Color Segmentation Through Morphological Model Refinement

This study used dynamic maps for skin color detection and applied a morphological closing operator to identify supplementary pixels in the image space to update the color map. Dual constraints were considered in the refinement system to prevent overestimation, which are the centroid of the skin color region in image space and dynamic range in color model space. As a result, the proposed system enables us to update the skin color map continuously. This paper presents a simple, rapid, and effective method for single skin color detection. This paper proposes skin color detection for images involving multiskin color regions (i.e., multiple human subjects of different races) by using the AdaBoost face detection mechanism. The experimental results reveal that the proposed method is highly robust for detection of diverse skin colors and simultaneous occurrences of multiskin color regions.

Latent class analysis of COVID-19 symptoms in Brazil: results of the PNAD-COVID19 survey.

The lack of mass testing for COVID-19 diagnosis creates the need to determine the magnitude of the disease based on its clinical symptoms. The study aimed to analyze the profile of COVID-19 symptoms and related aspects in Brazil. The author analyzed the sample of participants from the Brazilian National Household Sample Survey (PNAD-COVID19) conducted in May 2020. Latent class analysis (LCA) was performed with sociodemographic covariables and 11 symptoms reported by 346,181 individuals. Rao-Scott test and standardized residual analysis were used to measure the association with the pattern of health services use. Spatial scan analysis was performed to identify areas at risk of COVID-19 cases. LCA showed six classes of symptoms based on the pattern of answers by participants: (1) all the symptoms; (2) high prevalence of symptoms; (3) predominance of fever; (4) predominance of cough/sore throat; (5) mild symptoms with predominance of headache; and (6) absence of symptoms. Female sex, brown skin color, the North and Northeast regions of Brazil, and all three older age brackets showed stronger association with the class with all the symptoms (class 1). Most use of health services was also by this group of individuals, but with different profiles of use. Spatial analysis showed juxtaposition of this class with areas at greater risk of COVID-19. These finding underline the importance of investigating symptoms for the epidemiological identification of possible cases in a scenario with low population testing rates.

Preferred skin reproduction centres for different skin groups

Producing preferred skin colours is vital for the digital images on mobile phone manufacturers. Previous studies investigated the skin colours only in chromatic plane excluding lightness. A psychophysical experiment was conducted to determine preferred skin colour centres for different skin colour types on mobile displays in a darkened room. Ten facial images were selected for the experiment to cover different skin colour types (Caucasian, Oriental, South Asian and African). A set of 49 predetermined colour centres uniformly sampled within the skin colour ellipsoid in CIELAB colour space was used to morph skin colours of test images. Thirty observers from each of the 3 ethnic groups (Caucasian, Oriental and South Asian) participated in the experiment. The preferred skin colour centre and region in the form of ellipsoid for each skin group were reported. It was found that the preferred colour centres from different skin colour types were very similar except their lightness as expected, and were also quite similar between the observers from different ethnic groups.

Analysis of face detection based on skin color characteristic and AdaBoost algorithm

For the face detection method based on skin color feature and AdaBoost algorithm, if one of them is used to detect the face, it can also catch the face to a certain extent. However, the detection rate and an error rate of this single method in its detection experiment can’t achieve good results. Therefore, this paper combines the advantages of the two algorithms, combines the two approaches, and improves them. The main idea is to use the skin color features of face detection as pre-detection, and use the established skin color distribution Gaussian model to obtain candidate regions containing the skin color of the face, and then use a cascade classifier to detect the skin color regions. By using OpenCV and Visual Studio software, a lot of experimental statistics and analysis are carried out. The research shows that the improved algorithm is superior to the two algorithms in detection rate and false detection rate, and it can also achieve a good detection effect for the face in a complicated situation.

Real-time automatic tracking of hand motion in RGB videos using local feature SIFT

This paper proposes a method for real-time visual tracking of moving hand in RGB videos without any segmentation process and background subtraction. We have used YCgCr converted version of YCbCr colour space for a more compact representation of the initial region of moving hand and then local feature SIFT to detect and track hand simultaneously. YCgCr has a high tendency for skin colour accretion and can effectively discriminate between the skin and non-skin colour regions. The approach demonstrates that using local features (SIFT) of only active region reduces the computation as well as make the method free from the challenges of freedom factor of hand and thus the methodology can detect the hand of any shape and size without being affected by background conditions. In general, researchers avoid using a normal camera for applications based on hand tracking, as RGB images are sensitive to illumination. Our work exhibits that the combination of YCgCr and two-stage feature matching through SIFT algorithm is successful in tracking non-rigid objects with less computation. The methodology is further evaluated with Kalman tracking in hand gesture recognition and is also compared with contemporary works.

Real-time automatic tracking of hand motion in RGB videos using local feature SIFT

This paper proposes a method for real-time visual tracking of moving hand in RGB videos without any segmentation process and background subtraction. We have used YCgCr converted version of YCbCr colour space for a more compact representation of the initial region of moving hand and then local feature SIFT to detect and track hand simultaneously. YCgCr has a high tendency for skin colour accretion and can effectively discriminate between the skin and non-skin colour regions. The approach demonstrates that using local features (SIFT) of only active region reduces the computation as well as make the method free from the challenges of freedom factor of hand and thus the methodology can detect the hand of any shape and size without being affected by background conditions. In general, researchers avoid using a normal camera for applications based on hand tracking, as RGB images are sensitive to illumination. Our work exhibits that the combination of YCgCr and two-stage feature matching through SIFT algorithm is successful in tracking non-rigid objects with less computation. The methodology is further evaluated with Kalman tracking in hand gesture recognition and is also compared with contemporary works.

Multi-face detection based on improved YCbCr color space

Face recognition has become a popular means of recognition. It is required in image retrieval, video monitoring, entrance and exit control of military bases or other sensitive areas and other occasions. Face recognition has three key steps: face detection, feature extraction and feature matching. Face detection plays an important role in face recognition. Skin color segmentation is an important part of face detection. The traditional skin color modeling rules of YCbCr color space still have some defects, so we made some improvements. We changed the parameters of the conversion matrix used in the conversion from RGB space to YCbCr color space, so that the chosen skin color region is closer to the face skin color region. By this way, the system can detect the face most accurately.

RETRACTED ARTICLE: Application research of personnel attendance technology based on multimedia video data processing

At present, the access control system-based face recognition mostly use devices similar to fingerprint punchers for face collection and recognition. Its recognition accuracy is low, and it leads to frequent problems in personnel attendance. Based on this, this study combines multimedia image and video analysis technology to adopt multi-angle video collection for personnel attendance. At the same time, this study uses Shumate’s method for video preliminary processing, and uses histogram equalization processing to perform deep processing of the image. In addition, through the closed operation, the discrete points in the skin color region can be connected with the skin color region, so that the face region is more fully enriched, and the system is constructed on the basis of the technology. In our work, the average recognition rate has reached 94.40% when the number of samples is 5. Through comparative analysis, the algorithm of this study has certain practicality, which can provide theoretical reference for subsequent related research.

Adaptive Skin Color Detection through Iterative Illuminant Color Estimation and Conversion for Preferred Skin Color Reproduction

A robust skin color detection method using adaptive skin color modeling on YES color space is proposed. The skin colors are classified into three skin regions according to the Y value, and each region is modeled as the skin color in ES plane. By introducing an iterative estimation method of the illuminant color, we effectively remove the pseudo-skin colors that have been erroneously detected in skin color using the standard deviation of E-histogram. We also improve the skin color detection rate using a post-processing expansion mask that can include highlight and shadow pixels near the skin region as the skin color region.

Color Grading Systems to Classify Ripeness of Apple Mango Fruit

Harvesting of apple mangoes fruit that simultaneously causes not all apple mango fruits are harvested ripe. Farmers should sort the apple mango manually. So, harvesting in large quantities will take a long time. There are several obstacles to sort the apple mango fruit manually. It needs skills and experiences to sort the apple mangoes fruit and it needs many workers. In addition, when the apple mango is ripe and not immediately sorted, it will be too ripe and decayed, so cause losses. Previous research has already made the apple mango classification system mango level but required an expensive tool and take a long time to process one apple mango. This research proposes a new strategy to measure the ripeness level of apple mango fruit, especially apple mango at low cost, without the expensive device and takes a short time based on the apple mango skin color region using the feed-forward neural network. The test used 214 images consisting of 4 classes. Validation test using k-fold of 7, 9, and 11 with an average accuracy of 90.61%, 91.41%, and 90.82%. Highest accuracy with a value of 91.41% on k-fold 9 while at k-fold 11 accuracy is lower but has the least standard deviation value. A small standard deviation means that the accuracy is more stable than others

New face features to detect multiple faces in complex background

We propose in this paper a novel method for fast detecting faces even in the presence of constraints such as variation in illumination, human skin tone and facial expression, pose, and background (indoor or outdoor). Our system processes color images in a manner that would decrease the area of a face that must be scanned and for this, a parametric model based on Gaussian mixture models (GMM) applied to segmented regions of skin color. To select, relevant and minimum features from the faces candidates firstly, the variance based Haar-like features are extracted than merged with local binary patterns (LBP) features previously extracted. The resulting fused vectors construct Support Vector Machine database training to achieve a high detection rate. To verify the effectiveness of the proposed method, we carried out a serial of detailed experiments on three difficult face detection datasets (Caltech, BAO and UCD) which contain images featuring both single and multiple faces, presented in a variety of positions and featuring complex backgrounds, both indoor and outdoor. Experimental results have shown that our approach gives better results (91.04%) than those obtained by systems based on primitive Haar-like features and AdaBoost, providing a higher detection rate of 16.51%. Furthermore, the shorter detection time of our method is guaranteed by reducing the dimension of feature vectors and by limited search of faces on only the skin-detected regions and not on the entire image.

Pornographic video detection with MapReduce

In the big data era, the video data of social media increase rapidly. To detect and block pornographic videos, traditional pornographic image detection methods cannot be applied directly to large-scaled video data. For this purpose, a parallel computing network has been set up by a lot of cheap computers for massive pornographic video data detection. First, we propose a key-frame extraction algorithm based on inter-frame similarity. This algorithm only uses the local information of the video and can be dispatched to multiple computers for parallel processing. The results of key-frame extraction are persisted to the distributed file system. Next, in order to determine whether a video contains pornographic key-frames, we propose a discriminative multiple Gaussian mixture models to extract skin color regions and an active relevance feedback bootstrap algorithm to detect the face. Finally, the geometric characteristics of the body are used to determine whether the key-frame is a pornographic image, and according to the number of pornographic key-frame in the video to decide whether the video is pornographic or not. Compared with some existing methods, the detection accuracy has been greatly improved. Because of the proposed methods are processed in different computer nodes for parallel computing, the processing speed is only related to the scale of the video data and the number of the computers. In practical applications, it can meet demands only need to select enough computers according to the scale of the video data. In theory, it can be used for video data at any scale.

An Improved Otsu’s Thresholding Algorithm on Gesture Segmentation

In gesture recognition systems, segmenting gestures from complex background is the hardest and the most critical part. Gesture segmentation is the prerequisite of following image processing, and the result of segmentation has a direct influence on the result of gesture recognition. This paper proposed an algorithm of adaptive threshold gesture segmentation based on skin color. First of all, the image should be transformed from RGB color space to YCbCr color space. After eliminating luminance component Y, similarity graph of skin color will be obtained from the Gaussian model. Then Otsu adaptive threshold algorithm is used to carry out binary processing for the similarity graph of skin color. After the segmentation of skin color regions, the morphology method is used to process binary image for determining the location of hands. Experimental results show that the detailed segmentation of skin color using the dynamic-adaptive threshold can improve noise resistance and can produce better results.

Combining image and global pixel distribution model for skin colour segmentation

Skin colour detection in unconstrained and natural environment is a critical research problem. Skin regions show slightly different chrominance properties for different ambient conditions. Accuracy of detection is mostly affected either due to background colour similarity with the skin and/or poor illumination condition. To address this specific issue, a novel skin detection method is proposed by utilizing the information of pixel distribution in an image for a particular colour space. In our method, a local skin distribution model (LSDM) is derived from the image pixel distribution model using pixels from the facial region as reference. Finally, a combined skin distribution model is obtained by fusing the LSDM with the global skin colour distribution model. Subsequently, a dynamic region growing (DRG) method is proposed to allow the skin regions to grow dynamically. Our proposed DRG minimizes the overall detection error. Experimental results show that proposed skin detection method can more accurately segment out the skin-coloured regions. We obtained total detection error of 12.87% for a standard database. This improvement is due to the fact that both local and global skin colour information are used for skin segmentation.

A novel approach for face detection using hybrid skin color model

Human face detection has become an area of interest in various biometric applications such as crowd surveillance, human–computer interaction, and many security related areas. It is a major field of current research because there is no deterministic algorithm to find the face(s) in a given image. Face detection is challenging due to varying illumination conditions, pose variations, the complexity of noises and image backgrounds. In this paper, a localized approach for face detection based on skin color segmentation and facial features is proposed. Algorithm proficiently analyzes various skin color models such as RGB, YCbCr, HSV and their combinations for the skin color detection because skin segmentation decreases the computational complexity. The segmented face regions are further classified using a set of facial features such as eye-mouth hole detection, bounding box, and eccentricity ratio. This method is tested on four databases: Bao database contains 157 images, Muct database contains 751 images, FEI database contains 2800 images and GTAV database contains 1188 images. The algorithm achieves an average accuracy of $$97.85~\%$$ . Comparison with Viola Jones, face detection using skin color model, and fast face detection based on skin segmentation and facial features methods has also been done. Further, we apply this method to cartoon characters. Our method works with images with some conditions. First, input images are not monochrome images but color images. Second, the characters skin color has to be near real people skin color. Skin color region is extracted from the image and after this apply the facial features to locate faces in an image. The proposed approach for cartoon characters faces detection performed well and achieves a good accuracy of $$76.00~\%$$ .

Robust skin color-based moving object detection for video surveillance

Robust skin color-based moving object detection for video surveillance is proposed. The objective of the proposed algorithm is to detect and track the target under complex situations. The proposed framework comprises four stages, which include preprocessing, skin color-based feature detection, feature classification, and target localization and tracking. In the preprocessing stage, the input image frame is smoothed using averaging filter and transformed into YCrCb color space. In skin color detection, skin color regions are detected using Otsu’s method of global thresholding. In the feature classification, histograms of both skin and nonskin regions are constructed and the features are classified into foregrounds and backgrounds based on Bayesian skin color classifier. The foreground skin regions are localized by a connected component labeling process. Finally, the localized foreground skin regions are confirmed as a target by verifying the region properties, and nontarget regions are rejected using the Euler method. At last, the target is tracked by enclosing the bounding box around the target region in all video frames. The experiment was conducted on various publicly available data sets and the performance was evaluated with baseline methods. It evidently shows that the proposed algorithm works well against slowly varying illumination, target rotations, scaling, fast, and abrupt motion changes.

A real-time vision-based hand gesture interaction system for virtual EAST

The virtual Experimental Advanced Superconducting Tokamak device (VEAST) is a very complicated 3D model, to interact with which, the traditional interaction devices are limited and inefficient. However, with the development of human-computer interaction (HCI), the hand gesture interaction has become a much popular choice in recent years. In this paper, we propose a real-time vision-based hand gesture interaction system for VEAST. By using one web camera, we can use our bare hand to interact with VEAST at a certain distance, which proves to be more efficient and direct than mouse. The system is composed of four modules: initialization, hand gesture recognition, interaction control and system settings. The hand gesture recognition method is based on codebook (CB) background modeling and open finger counting. Firstly, we build a background model with CB algorithm. Then, we segment the hand region by detecting skin color regions with “elliptical boundary model” in CbCr flat of YCbCr color space. Open finger which is used as a key feature of gesture can be tracked by an improved curvature-based method. Based on the method, we define nine gestures for interaction control of VEAST. Finally, we design a test to demonstrate effectiveness of our system.

Faces detection method based on skin color modeling

This study suggests a method to improve the speed of a sliding window type of face detector by way of skin color region detection. The face detection method by way of skin color region detection has been studied in various perspectives: Complicated background images because of the area whose color is similar to the skin color cause high false positive rates. In contrast, the face detection method based on appearance, which adopts a sliding window type, may involve high face detection rates but cause tremendous computational costs in the process of detection scanning as the image size increases, whereas the processing time is also extended accordingly. This study suggests a method to control the subwindow size and detection area of a sliding window by detecting and using the skin color region with the processing time reduced. By means of a face detector with haar wavelet and LBP features, 274 images were collected online in addition to Bao database images, and then an experiment was conducted with them. As a result, the face detection time in utilization of an existing sliding window decreased down to a maximum of 47%.