Input Color Research Articles

Large-scale painting of photographs by interactive optimization

We propose a system for painting large-scale murals of arbitrary input photographs. To that end, we choose spray paint, which is easy to use and affordable, yet requires skill to create interesting murals. An untrained user simply waves a programmatically actuated spray can in front of the canvas. Our system tracks the can׳s position and determines the optimal amount of paint to disperse to best approximate the input image. We accurately calibrate our spray paint simulation model in a pre-process and devise optimization routines for run-time paint dispersal decisions. Our setup is light-weight: it includes two webcams and QR-coded cubes for tracking, and a small actuation device for the spray can, attached via a 3D-printed mount. The system performs at haptic rates, which allows the user – informed by a visualization of the image residual – to guide the system interactively to recover low frequency features. We validate our pipeline for a variety of grayscale and color input images and present results in simulation and physically realized murals.

Semi-parametric color reproduction method for optical see-through head-mounted displays.

The fundamental issues in Augmented Reality (AR) are on how to naturally mediate the reality with virtual content as seen by users. In AR applications with Optical See-Through Head-Mounted Displays (OST-HMD), the issues often raise the problem of rendering color on the OST-HMD consistently to input colors. However, due to various display constraints and eye properties, it is still a challenging task to indistinguishably reproduce the colors on OST-HMDs. An approach to solve this problem is to pre-process the input color so that a user perceives the output color on the display to be the same as the input. We propose a color calibration method for OST-HMDs. We start from modeling the physical optics in the rendering and perception process between the HMD and the eye. We treat the color distortion as a semi-parametric model which separates the non-linear color distortion and the linear color shift. We demonstrate that calibrated images regain their original appearance on two OST-HMD setups with both synthetic and real datasets. Furthermore, we analyze the limitations of the proposed method and remaining problems of the color reproduction in OST-HMDs. We then discuss how to realize more practical color reproduction methods for future HMD-eye system.

Iterative Refinement of Alpha Matte using Binary Classifier

Image matting is a technique used to extract a foreground object from an input color image by estimating its opacity mask, which is called an alpha matte. Many previous methods have estimated a poor-quality alpha matte when a foreground object and its background have similar colors. To overcome this problem, we treat an image matting problem as a binary classification problem to classify unknown pixels into foreground and background pixels. In this study, we propose a learning-based matting method based on binary classification. In our method, a binary alpha matte is first obtained by a binary classifier. The binary alpha matte is then iteratively refined to a high-quality alpha matte by repeat use of the classifier. Although we used a support vector machine classifier and the closed-form matting method, an important merit of our method is that it provides a general mechanism to refine an alpha matte by combining various binary classifiers and matting methods. The excellent performance of our method was revealed in several experiments, especially, for input images having similar foreground and background colors as well as objects with small holes.

Real-time expression transfer for facial reenactment

We present a method for the real-time transfer of facial expressions from an actor in a source video to an actor in a target video, thus enabling the ad-hoc control of the facial expressions of the target actor. The novelty of our approach lies in the transfer and photorealistic re-rendering of facial deformations and detail into the target video in a way that the newly-synthesized expressions are virtually indistinguishable from a real video. To achieve this, we accurately capture the facial performances of the source and target subjects in real-time using a commodity RGB-D sensor. For each frame, we jointly fit a parametric model for identity, expression, and skin reflectance to the input color and depth data, and also reconstruct the scene lighting. For expression transfer, we compute the difference between the source and target expressions in parameter space, and modify the target parameters to match the source expressions. A major challenge is the convincing re-rendering of the synthesized target face into the corresponding video stream. This requires a careful consideration of the lighting and shading design, which both must correspond to the real-world environment. We demonstrate our method in a live setup, where we modify a video conference feed such that the facial expressions of a different person (e.g., translator) are matched in real-time.

An Efficient Approach toward Color Image Segmentation with Combined Effort of Soft Clustering and Region based Techniques using L Channel of LAB Color Space

an efficient approach for color image segmentation is always sought by the researchers in the color image processing research. We have different clustering based and region based methods for the same. But still there arises the requirement of an optimal method. In this paper, a new approach for color image segmentation is proposed. Here the segmentation is carried out on the L channel of LAB color space. The input color image is first converted from RGB to LAB. Then L channel is extracted from the LAB converted image and sent as input to FCM algorithm. After this initial segmentation, the segmented image is filtered by sobel filter. The filtered image is then segmented by Meyer's Watershed algorithm to produce the final segmented image of the original image. The results of the proposed approach are found efficient when the same are analyzed in terms of MSE and PSNR. Also the segmented images are found free from over segmentation.

Car License Plate Detection

paper deals with car license plate detection (CLPD) system in order to identify vehicles by capturing their car license plates (CLP). Car license plate detection (CLPDS) is an emerging area of research due to various applications such as prevention of crime, electronic toll system, intelligent traffic control system etc. In the proposed system, after converting the color input image into grayscale, an adaptive thresholding is used to obtain the binary image. Then the unwanted lines are removed through an unwanted-line elimination algorithm (ULEA). Finally to detect the license plate, vertical edges are detected by Sobel operator. Experiments were carried out for detection of front view as well as rear view of license plates. The experimental evaluation is carried out for 60 images taken from roadside and parking lots. The proposed method yields 96.9% detection accuracy.

Efficient Light-Field Rendering Using Depth Maps for 100-Mpixel Multi-Projection 3D Display

In order to achieve an immersive, natural 3D experience on a large screen, a 100-Mpixel multi-projection 3D display was developed. Ninety-six projectors were used to increase the number of rays emanating from each pixel in the horizontal direction to 96. Conventional algorithms use a large number of cameras or input images to process a large number of light rays. This creates difficulties in the design of both the large acquiring system and substantial memory storage. In this paper, we propose an efficient light-field rendering algorithm that utilizes only a few input colors and depth images. Using a depth map and estimated camera parameters, synthesized light-field images are directly generated. This algorithm requires a much lighter memory load than conventional light-field rendering algorithms. It is also much simpler than the image-based rendering algorithm because it does not require the generation of so many multiview images.

Object knowledge changes visual appearance: Semantic effects on color afterimages

According to predictive coding models of perception, what we see is determined jointly by the current input and the priors established by previous experience, expectations, and other contextual factors. The same input can thus be perceived differently depending on the priors that are brought to bear during viewing. Here, I show that expected (diagnostic) colors are perceived more vividly than arbitrary or unexpected colors, particularly when color input is unreliable. Participants were tested on a version of the ‘Spanish Castle Illusion’ in which viewing a hue-inverted image renders a subsequently shown achromatic version of the image in vivid color. Adapting to objects with intrinsic colors (e.g., a pumpkin) led to stronger afterimages than adapting to arbitrarily colored objects (e.g., a pumpkin-colored car). Considerably stronger afterimages were also produced by scenes containing intrinsically colored elements (grass, sky) compared to scenes with arbitrarily colored objects (books). The differences between images with diagnostic and arbitrary colors disappeared when the association between the image and color priors was weakened by, e.g., presenting the image upside-down, consistent with the prediction that color appearance is being modulated by color knowledge. Visual inputs that conflict with prior knowledge appear to be phenomenologically discounted, but this discounting is moderated by input certainty, as shown by the final study which uses conventional images rather than afterimages. As input certainty is increased, unexpected colors can become easier to detect than expected ones, a result consistent with predictive-coding models.

Statistical convergence behavior of affine projection algorithms

Class of algorithms referring to the affine projection algorithms (APA) applies updates to the weights in a direction that is orthogonal to the most recent input vectors. This speeds up the convergence of the algorithm over that of the normalized least mean square (NLMS) algorithm, especially for highly colored input processes. In this paper a new statistical analysis model is used to analyze the APA class of algorithms with unity step size. Four assumptions are made, which are based on the direction vector for the APA class. Under these assumptions, deterministic recursive equations for the weight error and for the mean-square error are derived. We also analyze the steady-state behavior of the APA class. The new model is applicable to input processes that are autoregressive as well as autoregressive-moving average, and therefore is useful under more general conditions than previous models for prediction of the mean square error of the APA class. Simulation results are provided to corroborate the analytical results.

A robust band-dependent variable step size NSAF algorithm against impulsive noises

Proposed is a new subband adaptive filter (SAF) algorithm by minimizing Huber’s cost function that is robust to impulsive noises. Generally, this algorithm works in the mode of the normalized SAF (NSAF) algorithm, while it behaves like the sign SAF (SSAF) algorithm only when the impulsive noises appear. To further improve the robustness of this algorithm against impulsive noises, the subband cutoff parameters are updated in a recursive way. Moreover, the proposed algorithm can be interpreted as a variable step size NSAF algorithm, thus it exhibits faster convergence rate and lower steady-state error than the NSAF. Simulation results, using different colored input signals in both impulsive and free-impulsive noise environments, show that the proposed algorithm works better than some existing algorithms.

Shading-based refinement on volumetric signed distance functions

We present a novel method to obtain fine-scale detail in 3D reconstructions generated with low-budget RGB-D cameras or other commodity scanning devices. As the depth data of these sensors is noisy, truncated signed distance fields are typically used to regularize out the noise, which unfortunately leads to over-smoothed results. In our approach, we leverage RGB data to refine these reconstructions through shading cues, as color input is typically of much higher resolution than the depth data. As a result, we obtain reconstructions with high geometric detail, far beyond the depth resolution of the camera itself. Our core contribution is shading-based refinement directly on the implicit surface representation, which is generated from globally-aligned RGB-D images. We formulate the inverse shading problem on the volumetric distance field, and present a novel objective function which jointly optimizes for fine-scale surface geometry and spatially-varying surface reflectance. In order to enable the efficient reconstruction of sub-millimeter detail, we store and process our surface using a sparse voxel hashing scheme which we augment by introducing a grid hierarchy. A tailored GPU-based Gauss-Newton solver enables us to refine large shape models to previously unseen resolution within only a few seconds.

A New Normalized Subband Adaptive Filter Algorithm with Individual Variable Step Sizes

Proposed is a novel variable step size normalized subband adaptive filter algorithm, which assigns an individual step size for each subband by minimizing the mean square of the noise-free a posterior subband error. Furthermore, a noniterative shrinkage method is used to recover the noise-free priori subband error from the noisy subband error signal. Simulation results using the colored input signals have demonstrated that the proposed algorithm not only has better tracking capability than the existing subband adaptive filter algorithms, but also exhibits lower steady-state error.

Hardware‐Based Non‐Photorealistic Rendering Using a Painting Robot

AbstractWe describe a painting machine and associated algorithms. Our modified industrial robot works with visual feedback and applies acrylic paint from a repository to a canvas until the created painting resembles a given input image or scene. The color differences between canvas and input are used to direct the application of new strokes. We present two optimization‐based algorithms that place such strokes in relation to already existing ones. Using these methods we are able to create different painting styles, one that tries to match the input colors with almost transparent strokes and another one that creates dithering patterns of opaque strokes that approximate the input color. The machine produces paintings that mimic those created by human painters and allows us to study the painting process as well as the creation of artworks.

A novel set of image morphological operators using a modified vector distance measure with color pixel classification

A novel set of color image morphological operators is proposed and is based on using a modified vector distance measure with an efficient preprocessing color pixel classification procedure. The input color pixels are classified into different categories based on the characteristics of the human color perception properties. Next, the color distance is calculated for each input color pixel based on its category, and, a combined technique is used to obtain color morphological operators. The main contributions of this work can be summarized in the following: (a) designing of novel vector-based morphological operators that can be used with color images and (b) the designed operators were extended to be implemented for developing vector-based tools for noise suppression, texture analysis, shape analysis, edge detection and skeletonization.

능동 소음 제어를 위한 적응 알고리즘들 비교

In this paper, we confirm the effective adaptive algorithm for tha active noise contorl (ANC) though the performance comparison between adaptive algorithms. Generally, the normalized least mean square (NLMS) algorithm has been widely used for an adaptive algorithm thanks to its simplicity and having a fast convergence speed. However, the convergence performance of the NLMS algorithms is often deteriorated by colored input signals. To overcome this problem, the affine pojection (AP) algorithm that updates the weight vector based on a number of recent input vectors can be used for allowing a higher convergence speed than the NLMS algorithm, but it is computationally complex. Thus, the proper algorithm were determined by the comparison between NLMS and AP algorithms regarding as the convergence performance and complexity. Simulation results confirmed that the noise reduction performance of NLMS algorithm was comparable to AP algorithm with low complexity. Therefore the NLMS algorithm is more effective for ANC system.

Detection of Bird Nests on Equipments of Electric Transmission Lines Based on Aerial Images

This paper presents a method for detecting bird nests on the equipments of electric transmission lines by aerial images. Firstly, an input color aerial image is transformed from the RGB space to the HSV space, and then bird nests can be segmented roughly from the aerial images by the K-means clustering algorithm. After that, some noises are removed by morphological filtering operations. Finally, the location of the bird nests in aerial images can be obtained by judging the shapes of the connected regions. Experiments show that this method is effective and can be applied to the helicopter patrol inspection system.

Convergence characteristics of a stereophonic acoustic echo canceler with a colored input signal

Convergence characteristics of a stereophonic acoustic echo canceler with a colored input signal

Gemini Virus Attack Analysis in Field of Chili (Capsicum Annuum L.) Using Aerial Photography and Bayesian Segmentation Method

One of the constraints of low productivity of chili caused by a viral infection called Gemini virus. Viral infections potentially reduce yields by 20-100%. Due to these conditions, it is necessary to look for methods in monitoring and evaluating virus attack on chili to get a better yields production. The purpose of this study is to analyze the chili farms that were exposed by virus. Image acquisition is done through aerial photography via multi-copter. Bayesian segmentation is used to determine the level of virus attacks on individual plant with a 3-dimensional input color component (R, G, B) and 4 segmentation target.

Color Image Segmentation Using Adaptive GrowCut Method

An adaptive algorithm for interactive multi-label segmentation of 2-dimensional color images using GrowCut has been carry out in this technique. Except certain small number of user-labelled pixels, the remainder of the image is segmented automatically by a layered automaton. This technique is iterative, since the mechanism labels the image automatically. The segmentation task is complex to compute. User can monitor the progress of segmentation and redirect the algorithm with respect to human participation. The proposed system contains two phases. During the first level the original input color image is segmented using the GrowCut method. In the following phase the resulting image is processed under color image segmentation method. In the areas, where the segmentation is dependably computed automatically no further user effort is necessary. Outcomes of segmenting standard photos and images are presented. Our experiments show that a little user effort is required for segmentation of moderately tough images.

ROBUST POTATO COLOR IMAGE SEGMENTATION USING ADAPTIVE FUZZY INFERENCE SYSTEM

Potato image segmentation is an important part of image-based potato defect detection. This paper presents a robust potato color image segmentation through a combination of a fuzzy rule based system, an image thresholding based on Genetic Algorithm (GA) optimization and morphologi- cal operators. The proposed potato color image segmentation is robust against variation of background, distance and view of potato from digital camera. In the proposed algorithm, after selecting appropriate color space, distance be- tween an image pixel and real potato pixels is computed. Furthermore, this distance feeds to a fuzzy rule-based classier to extract potato candidate in the input image. A subtractive clustering algorithm is also used to decide on the number of rules and membership functions of the fuzzy system. To improve the performance of the fuzzy rule-based classier, the membership functions shapes are also optimized by the GA. To segment potatoes in the input color image, an image thresholding is applied to the output of the fuzzy system, where the corresponding threshold is optimized by the GA. To im- prove the segmentation results, a sequence of some morphological operators are also applied to the output of thresholding stage. The proposed algorithm is applied to dierent databases with dierent backgrounds, including USDA, CFIA, and obtained potato images database from Ardabil (Iran's northwest), separately. The correct segmentation rate of the proposed algorithm is approx- imately 98% over totally more than 500 potato images. Finally, the results of the proposed segmentation algorithm are evaluated for some images taken from real environments of potato industries and farms.