Mihai Ciuc

Adaptive-neighborhood histogram equalization of color images

Histogram equalization (HE) is one of the simplest and most effective techniques for enhancing gray-level images. For color images, HE becomes a more difficult task, due to the vectorial nature of the data. We propose a new method for color image enhancement that uses two hierarchical levels of HE: global and local. In order to preserve the hue, equalization is only applied to intensities. For each pixel (called the ‘‘seed’’ when being processed) a variable-sized, variable-shaped neighborhood is determined to contain pixels that are ‘‘similar’’ to the seed. Then, the histogram of the region is stretched to a range that is computed with respect to the statistical parameters of the region (mean and variance) and to the global HE function (of intensities), and only the seed pixel is given a new intensity value. We applied the proposed color HE method to various images and observed the results to be subjectively ‘‘pleasant to the human eye,’’ with emphasized details, preserved colors, and with the histogram of intensities close to the ideal uniform one. The results compared favorably with those of three other methods (histogram explosion, histogram decimation, and three-dimensional histogram equalization) in terms of subjective visual quality.

Measurements with ToF Cameras and Their Necessary Corrections

The most important characteristic of time-of-flight (ToF) cameras is the ability to measure the distance to each image pixel. Thus, for each pixel, information on both its amplitude and distance to the camera are available. However, technological problems inherent to the acquisition principle lead to inaccuracies in estimating both characteristics: on one hand, there are errors in estimating the distance, especially for far-distance pixels. On the other hand, the detected amplitude decreases with the distance. Part of these inaccuracies are corrected with special camera-calibration software. In this paper, we propose two methods that attempt to further correct each information based on the other one. First, the amplitude image is enhanced by using distance information: a pixel-wise, distance-based correction of the amplitude brings to light details otherwise unnoticeable. Secondly, an amplitude-based distance modification corrects some of the distance estimation errors for far-distance pixels.

General adaptive-neighborhood technique for improving synthetic aperture radar interferometric coherence estimation

A new method for filtering the coherence map issued from synthetic aperture radar (SAR) interferometric data is presented. For each pixel of the interferogram, an adaptive neighborhood is determined by a region-growing technique driven by the information provided by the amplitude images. Then pixels in the derived adaptive neighborhood are complex averaged to yield the filtered value of the coherence, after a phase-compensation step is performed. An extension of the algorithm is proposed for polarimetric interferometric SAR images. The proposed method has been applied to both European Remote Sensing (ERS) satellite SAR images and airborne high-resolution polarimetric interferometric SAR images. Both subjective and objective performance analysis, including coherence edge detection, shows that the proposed method provides better results than the standard phase-compensated fixed multilook filter and the Lee adaptive coherence filter.

Intensity-driven-adaptive-neighborhood technique for POLSAR parameters estimation

In this paper, a new method to estimate polarimetric coherency matrices and derive associated parameters is presented. For each pixel of the data set, an adaptive neighborhood is computed by a region growing technique driven exclusively by the intensity images. The three intensity images of the POLSAR acquisition are fused in the region growing process to ensure the stationarity hypothesis of the derived statistical population. Then, all pixels within the obtained adaptive neighborhood are, either complex averaged or estimated by the locally linear minimum mean squared error (LLMMSE), to yield a feature preserving reliable estimate of the polarimetric coherency matrix. The target entropy/alpha/anisotropy decomposition is applied on the derived polarimetric coherency matrix. Using this decomposition, unsupervised classifcation for land applications by an iterative algorithm based on a complex Wishart density function is employed. The method has been tested on airborne polarimetric synthetic aperture radar images (Northumberland Strait costal area - Canadian Space Agency).

Detection of pectoral muscle in mammograms using a mean-shift segmentation approach

Detection of pectoral muscle in mammograms is an important pre-processing segmentation step. The pectoral muscle is one of the few anatomical features that appears clearly and reliably in medio-lateral oblique view mammograms. This new method overcomes the limitation of the straight-line representation considered in our initial investigation using the Hough transform.

Facial enhancement and beautification for HD video cameras

Influenced by the widespread adoption of HDTV consumer imaging devices have begun to feature full HD (high density) video capture. With this high-resolution capability, small imperfections in human faces are captured with full HD video. When used for portrait imaging, the resulting video is frequently unsatisfactory to the user. Manufacturers of imaging devices require practical, real-time solutions which can mitigate facial details emphasized by the HD resolution, while preserving the overall quality of the portrait images. The high bandwidth and processing requirements of full HD video make this a challenging task. In this paper, a practical algorithm, now incorporated in a number of consumer devices is explained. Major challenges and their solutions are presented. Attention is given to real-time optimizations of the algorithm which is designed with a view to its suitability for partial or full hardware implementation.

Normalized Autobinomial Markov Channels For Pedestrian Detection.

Pedestrian detection represents one of the most important components of engineering devices that use automated vision to help decision systems take quick and accurate actions. Such systems are defined and customized to be useful for different needs, such as monitoring and aided surveillance, or increasing safety features in automotive industry. Given the large spectrum of applications that use pedestrian detection, demand has increased in recent years for the development of feasible solutions which can be integrated in devices such as smartphones or action cameras. This paper focuses on finding probabilistic features that highlight the human body characteristics regardless of contextual information in images. Adjacent pixels are often spatially correlated, which means that they are likely to have similar values. We view the image as a collection of random variables indexed by certain locations, called sites. The state of a site ξ is conditionally independent of all variables in the random field, except the neighbouring system Nξ = { η ∈Ω | η 6= ξ , d2(ξ ,η)≤ ∆ } , where ∆ is a positive integer and d2(ξ ,η) is the squared Euclidean distance between ξ and η . The neighbouring system strictly depends on a collection of cliques C = ∑ ) k=1 Ck, where ω(∆) is the number of cliques for each local specification. Energy function: An unpublished manuscript [2] describes how to interpret the local property of a Markov random field in terms of energy and potential, claiming that the probability at a site ξ is given by:

A new approach in breast image registration

According to the World Health Organization, breast cancer is the most common cancer suffered by women in the world, which during the last two decades, has increased the women mortality in developing countries. Mammography is the best method used for the screening; the problem of detecting possible cancer areas is very complex due, on one hand, to the diversity in shape of the ill tissue and, on the other hand, to the poorly defined border between the healthy and the cancerous zone. An automated technique for the alignment of right and left breast images has been developed for use in the computerized analysis of bilateral breast images. Using this technique, the breast region is firstly identified by using an adaptive thresholding algorithm. The focus is on determining control points in the two mamograms; these points are used to put the two mammograms into correspondence. The algorithms performance was evaluated on a large number of difficult cases and found to be adequate.

Automatic pap smear nuclei detection using mean-shift and region growing

The Babes-Papanicolaou test (also known as Pap smear) is a method of cervical cancer screening used to detect abnormal cells which are or can become cancerous. Since the visual inspection of pap smears is very time consuming, the need for automatic methods is required. This paper presents an algorithm for the automatic detection of nuclei within pap smears images. The algorithm relies in the highly effective mean-shift filtering method which enhances the contrast of nuclei areas. The segmentation consists of a region growing with starting points taken from the image gradient map. Size and eccentricity measures are used to keep only nuclei from the segmented regions. The method is validated on two different pap smear test databases and the detection rate is above 91%.

Digital Beauty: The good, the bad, and the (not-so) ugly.

Today, digital retouching of your pictures is made possible in the latest smartphones and cameras at the touch of a button. What is more, all of this can be achieved transparently to the user, in real time, just as the image is acquired or added afterwards, allowing users to manipulate and enhance individual faces according to their personal preferences.