Original Gray-level Images Research Articles

Facial Expression Recognition based on Image Gradient and Deep Convolutional Neural Network

Facial Expression Recognition (FER) is one of the basic ways of interacting with machines and has been getting more attention in recent years. In this paper, a novel FER system based on a deep convolutional neural network (DCNN) is presented. Motivated by the powerful ability of DCNN to learn features and image classification, the goal of this research is to design a compatible and discriminative input for pre-trained AlexNet-DCNN. The proposed method consists of 4 steps: first, extracting three channels of the image including the original gray-level image, in addition to horizontal and vertical gradients of the image similar to the red, green, and blue color channels of an RGB image as the DCNN input. Second, data augmentation including scale, rotation, width shift, height shift, zoom, horizontal flip, and vertical flip of the images are prepared in addition to the original images for training the DCNN. Then, the AlexNet-DCNN model is applied to learn high-level features corresponding to different emotion classes. Finally, transfer learning is implemented on the proposed model and the presented model is fine-tuned on target datasets. The average recognition accuracy of 92.41% and 93.66% were achieved for JAFEE and CK+ datasets, respectively. Experimental results on two benchmark emotional datasets show promising performance of the proposed model that can improve the performance of current FER systems.

Binarization of uneven-lighting image by maximizing boundary connectivity

By using the constraint of boundary connectivity, a new technique for binarizing a gray level image is proposed. Initially, the gray level image is partitioned into a number of fixed-size subimages and the best binarization of each subimage is estimated. Of all subimages, the one with the best binarization effect is first binarized to be the starting binary area of the gray level image. Next, of the subimages neighboring the binary area, the one with the best binarization effect is selected to perform the following test. If the maximal boundary connectivity achieved by binarizing the selected subimage with a specific threshold satisfies the constraint of boundary connectivity, the subimage will thus be binarized to update the partially-binarized gray level image, otherwise, the subimage will be marked “ignored” and remain gray throughout the subimage binarization procedure. The procedure will be repeated until no further subimages neighboring the binary area can be binarized. If there are any gray subimages without being marked “ignored,” the one with the best binarization effect will be binarized to create a new binary area in the partially-binarized gray level image, and then the procedure of extending the binary area will be repeated again. At the last stage, the proposed method constructs a threshold surface for the original gray level image based on the binarized subimages. Experimental results show the effectiveness of this method in binarizing uneven-lighting gray level images.

Image Compression Using Modified Huffman Coding

Image compression using modified Huffman coding can be obtained by dividing the original grey level image into unoverlapped blocks and setting homogeneity criterion to test each block individually depending on a threshold value and the well-known techniques of Quad-tree decomposition. By taking threshold value as 0.2, Huffman coding for encoding and decoding of the image are the techniques which have been applied for the compression of satellite imageries. The results indicate that for a Lena image, CR is 9.79 and PSNR value is 24.96, Satellite image values are 12.69 and 25.01, respectively. The noise used is salt and pepper, and it is filtered by wellknown technique of Wiener filter. The results are presented and discussed in this paper.

Face recognition via edge-based Gabor feature representation for plastic surgery-altered images

Plastic surgery procedures on the face introduce skin texture variations between images of the same person (intra-subject), thereby making the task of face recognition more difficult than in normal scenario. Usually, in contemporary face recognition systems, the original gray-level face image is used as input to the Gabor descriptor, which translates to encoding some texture properties of the face image. The texture-encoding process significantly degrades the performance of such systems in the case of plastic surgery due to the presence of surgically induced intra-subject variations. Based on the proposition that the shape of significant facial components such as eyes, nose, eyebrow, and mouth remains unchanged after plastic surgery, this paper employs an edge-based Gabor feature representation approach for the recognition of surgically altered face images. We use the edge information, which is dependent on the shapes of the significant facial components, to address the plastic surgery-induced texture variation problems. To ensure that the significant facial components represent useful edge information with little or no false edges, a simple illumination normalization technique is proposed for preprocessing. Gabor wavelet is applied to the edge image to accentuate on the uniqueness of the significant facial components for discriminating among different subjects. The performance of the proposed method is evaluated on the Georgia Tech (GT) and the Labeled Faces in the Wild (LFW) databases with illumination and expression problems, and the plastic surgery database with texture changes. Results show that the proposed edge-based Gabor feature representation approach is robust against plastic surgery-induced face variations amidst expression and illumination problems and outperforms the existing plastic surgery face recognition methods reported in the literature.

QPSK Modulation Based Optical Image Cryptosystem Using Phase-shifting Digital Holography

We propose a new technique for the optical encryption of gray-level optical images digitized into 8-bits binary data by ASCII encoding followed by QPSK modulation. We made an encrypted digital hologram with a security key by using 2-step phase-shifting digital holography, and the encrypted digital hologram is recorded on a CCD camera with 256 gray-level quantized intensities. With these encrypted digital holograms, the phase values are reconstructed by the same security key and are decrypted into the original gray-level optical image by demodulation and decoding. Simulation results show that the proposed method can be used for cryptosystems and security systems.

Image secret sharing method with two-decoding-options: Lossless recovery and previewing capability

Visual cryptography scheme (VCS) is a research area in image secret sharing, where one can easily stack shadow images and decode a black-and-white secret image through the human visual system without computation. Although VCS does not provide a competitive reconstruction quality, its stacking-to-see property clearly has the intended applications. To develop the ease of decoding of VCS and simultaneously overcome its weakness, Lin et al. recently proposed a novel two-in-one image secret sharing scheme (TiOISSS) with two decoding options based on VCS and a polynomial-based image secret sharing scheme (PISSS). In this TiOISSS, the first decoding stage has the stacking-to-see property to preview a vague image, and the second decoding stage is to obtain the original gray-level secret image. In this paper, we design a new TiOISSS by combining VCS and PISSS in a different way; in specific, our new scheme reduce shadow image size, and is more suitable for faster transmission within a distributed multimedia system.

Inverse error-diffusion using iterated conditional modes

Error-diffusion (ED) is one conventional halftoning technique that converts a gray-level image into a halftone. For further processing the ED halftones, it is often necessary to estimate the original image from the halftone: the inverse of ED. We propose to calculate iterated conditional modes (ICM) for the maximum a posteriori (MAP) solution of inverse ED. The ICM always searches for a better estimate in the valid image space. It requires only local computation and is applicable to any type of the MRF model used for the original gray-level images. Experimental results for common standard images are given to show that our ICM performs well and is more flexible than the descent-projection (DP) approach.

Analyzing Protein-Protein Spatial-Temporal Dependencies from Image Sequences Using Fuzzy Temporal Random Sets

Total Internal Reflection Fluorescence Microscopy (TIRFM) allows us to image fluorescenttagged proteins near the plasma membrane of living cells with high spatial-temporal resolution. Using TIRFM imaging of GFP-tagged clathrin endocytic proteins, areas of fluorescence are observed as overlapping spots of different sizes and durations. Standard procedures to measure protein-protein colocalization of dual labeled samples threshold the original graylevel images to segment areas covered by different proteins. This binary logic is not appropriate as it leaves a free tuning parameter which can influence the conclusions. Moreover, these procedures rely on simple statistical analysis based on correlation coefficients or visual inspection. We propose a probabilistic model to examine spatial-temporal dependencies. Image sequences of two proteins are modeled as a realization of a bivariate fuzzy temporal random set. Spatial-temporal dependencies are described by means of the pair-correlation function and the K-function and are tested using a Monte Carlo test. Five simulated image sequences were used to validate the performance of the procedure. Spatial and spatial-temporal dependencies were generated using a linked pairs model and a Poisson cluster model for the germs. To demonstrate the applicability in addressing current biological questions, we applied the procedure to fluorescent-tagged proteins involved in endocytosis (Clathrin, Hip1R, Epsin, and Caveolin). Results show that this procedure allows biologists to automatically quantify dependencies between molecules in a more formal and robust way. Image sequences and a Matlab toolbox for simulation and testing are available at http://www.uv.es/tracs/.

Edge Detection on Hexagonal Structure

Gradient-based edge detection is a straightforward method to identify the edge points in the original grey-level image. It is intuitive that in the human vision system the edge points always appear where the grey-level value is greatly changed. Spiral Architecture is a relatively new image data structure that is inspired from anatomical considerations of the primate's vision. In Spiral Architecture, each image is represented as a collection of hexagonal pixels. Edge detection on Spiral Architecture has features of fast computation and accurate localization. In this paper, we present and compare gradient-based edge detection algorithms on Spiral Architecture. The experimental results show that the edge detection on Spiral Architecture outperforms that on traditional square image structure.

Gradient-based edge detection on a hexagonal structure

Gradient-based edge detection is a straightforward method to identify the edge points in the original grey-level image. It is intuitive that in the human vision system the edge points always appear where the grey-level value is greatly changed. Spiral Architecture is a relatively new image data structure that is inspired from anatomical considerations of the primate’s vision. In Spiral Architecture, each image is represented as a collection of hexagonal pixels. Edge detection on Spiral Architecture has features of fast computation and accurate localization. In this paper, we review the gradient-based edge detection algorithms on Spiral Architecture. An edge point is defined as a hexagonal pixel at which the magnitude of the gradient of brightness function assumes a local maximum.

A New Model-Based Digital Halftoning and Data Hiding Designed With LMS Optimization

This work employs the well known least-mean-square (LMS) method to design an adaptive filter to produce high-quality halftone images. The filter can be regarded as a transformation medium between original gray level images and corresponding halftone images. Experimental results indicate that the proposed LMS-designed halftoning offers the extra benefit of edge enhancement. Since a halftone image is typically used in printing, a modified printer model, which can coordinate with the proposed LMS-designed halftoning, is proposed to eliminate the harm caused by the dot-gain effect in printing. Moreover, two data hiding applications, the direct embedding LMS-designed halftone technique (DELDH) and the information sharing LMS-designed halftone technique (ISLDH), are proposed to demonstrate the performance of the proposed LMS-designed halftoning. The experimental results show that, both techniques can be used with the proposed modified printer model to achieve excellent image quality and decoded visual patterns

A reinforcement agent for threshold fusion

Finding an optimal threshold in order to segment digital images is a difficult task in image processing. Numerous approaches to image thresholding already exist in the literature. In this work, a reinforced threshold fusion for image binarization is introduced which aggregates existing thresholding techniques. The reinforcement agent learns the optimal weights for different thresholds and segments the image globally. A fuzzy reward function is employed to measure object similarities between the binarized image and the original gray-level image, and provide feedback to the agent. The experiments show that promising improvement can be obtained. Three well-established thresholding techniques are combined by the reinforcement agent and the results are compared using error measurements based on ground-truth images.

A new progressive coding scheme for halftone dithered images using bit-interleaving

A novel progressive coding scheme is presented for the efficient display of dithered images. Dithered images are the results of thresholding original gray-level images with dithering screens. After the preprocessing of bit-interleaving, this algorithm utilizes the characteristic of reordered image to determine the transmitting order and then progressively reconstructs the dithered image. In addition, the dithered images are further compressed by lossy and lossless procedures. The experimental results demonstrate high-quality reconstructions while maintaining low transmitted bit rates. Moreover, an objective error criterion obtained by LMS optimization is also presented in this study. As documented in the experiments, the error criterion is quite consistent to the basic characteristic of human visual system. © 2007 Wiley Periodicals, Inc. Int J Imaging Syst Technol, 16, 121–127, 2006.

Data hiding in halftone images with noise-balanced error diffusion

In this letter, we propose a low-complexity algorithm for embedding watermarks into two or more error-diffused images. The first one is only a regular error-diffused image, and the others are achieved by applying the proposed noise-balanced error diffusion technique (NBEDF) to the original gray-level image. The visual decoding pattern can be perceived when these two or more similar error-diffused images are overlaid each other. Furthermore, with the proposed modified version of NBEDF, the two halftone images can be made from two totally different gray-tone images and still provide a clear and sharp visual decoding pattern.

고속 검사자동화를 위한 에지기반 점 상관 알고리즘의 개발

We presents an edge-based point correlation algorithm for fast and stable visual inspection system. Conventional algorithms based on NGC(normalized gray-level correlation) have to overcome some difficulties in applying automated inspection systems to real factory environment. First of all, NGC algorithms involve highly complex computation and thus require high performance hardware for realtime process. In addition, lighting condition in realistic factory environments is not stable and therefore intensity variation from uncontrolled lights gives many troubles for applying NGC directly as pattern matching algorithm. We propose an algorithm to solve these problems, using thinned and binarized edge data, which are obtained from the original image. A point correlation algorithm with the thinned edges is introduced with image pyramid technique to reduce the computational complexity. Matching edges instead of using original gray-level image pixels overcomes problems in NGC method and pyramid of edges also provides fast and stable processing. All proposed methods are proved by the experiments using real images.

Image analysis characterization of multi-walled carbon nanotubes

An original image analysis method is presented to characterize multi-walled carbon nanotubes from transmission electron microscopy images. The analysis is performed in three steps: (i) image preprocessing in order to isolate the nanotubes from the background, (ii) image segmentation, aiming at keeping only the measurable sections of nanotubes, and finally (iii) tube characteristics measurement. The measurement is based on a Lambert-like electron absorption law and is performed on the original gray level image itself. Two geometrical and one physical characteristics are determined for each tube, namely, its outer and inner radius and a linear electron absorption coefficient. The method is illustrated by comparing a pristine and an annealed carbon nanotube samples. The compaction of the tube walls during annealing is shown to result from a lowering of the external radius while the inner radius is left unchanged.

A new optimal digital halftoning technique based on the discrete cosine transform

Digital halftoning is a technique to display a gray-level image with a bilevel device. Conventionally, most halftoning techniques are done in the spatial domain. A new halftoning technique based on the discrete cosine transform is proposed. The method chooses an optimal bilevel image to display the original gray-level image and minimize the weighted mean square error based on the discrete cosine transform domain. The simulation results indicate that our algorithm can produce very good halftoned images without false contours.

A Study Of Diffusion Zones With Electron Microprobe Compositional Mapping

Digital compositional mapping with the electron microprobe has been used to study diffusion zones at metal-metal and metal-glass interfaces. Image enhancement techniques and concentration-histogram images (CHI’s) have been used with these digital maps to improve visual display and interpretation of the movement of elements in these zones.Metal-Metal Interface. In figure la is a WDS concentration map of gold taken from a specimen of a gold sheet about 40 μm wide sandwiched between two platinum sheets. This specimen was observed as part of a metallurgical study of some ancient Ecuadorian artifacts to better understand the original reaction/diffusion zone between these two metals. The specimen was annealed at 1350°C for 3 min. and then rapidly cooled. This temperature was sufficient to melt the gold (M.P. 1064°C) but not the platinum (M.P. 1769°C). The map in figure la shows that some platinum has diffused into the gold region, but the extent of diffusion is difficult to see in this original grey-level image.

Unsupervised classification of cell images using pyramid node linking

In this communication we describe a segmentation technique which combines two properties in an iterative and hierarchial matter to correctly segment and classify the given cell images. The technique is applied to digital images taken from microscope slides of cultured rat liver cells, and the goal is to classify these cells into one of three possible classes. The first class cells (I) are morphologically normal and stain the darkest. The second class cells (II) are slightly damaged showing both nuclear and cytoplasmic swelling with resultant lessening of staining affinity. The third class cells (III) are markedly damaged as demonstrated by the presence of cytoplasmic vacuolization, or are completely disintegrated. First class cells are classified by taking advantage of their staining affinity; the original gray level image is segmented into four gray levels. The darkest is then classified as type I. Type III cells are classified by using high business as a characteristic; the standard deviation of the original image is segmented into four business levels. The highest level is classified as type III cell. Assuming only the three cell types are present in any given image, the remaining non-background unclassified pixels are determined to belong to type II cells.

Projection-based high accuracy measurement of straight line edges

We present a novel projection-based high accuracy algorithm to determine the parameters of a straight edge in a noisy image. Our algorithm is equivalent to applying a set of very long directional operators over a range of finely quantized angles. This is known to improve both signal-to-noise ratio and localization in the measurement of straight line edges, but was rarely used in practice because of the expense and coefficient restrictions of the operators. The algorithm is implemented by taking a set of projections of the original grey level image, filtering the projections, and analyzing the peaks in projection space to estimate line offset and angle. It also includes a procedure to analyze the peaks in projection space which provides an accuracy better than the quantization in offset and angle parameters. The algorithm was tested on synthetic and a large number of real images and offers very high (subpixel) offset and angular accuracy. The advantages of the algorithm over traditional approaches are improved signal-to-noise ratio and localization accuracy (due to the effective use of very long directional edge operators), no need for expensive edge operators and related hardware, and no need for sophisticated thresholding of the gradient image for finding edges. It is robust in the presence of many types of texture, patterned and bias noise, light intensity, and focus change, and is ideal for use in industrial machine vision, where a large number of parts with straight edges are processed.