A frequent degradation in video-guided surgery and especially laparoscopic and endoscopic surgery is uneven illumination, due in large part to physical limitations of the sensors and uncontrolled lighting conditions in the internal structure of the digestive tract and particularly at the level of the intestines. Surgical as well as postoperative task accuracy can be seriously affected by the perceptual quality of the acquired images or videos. In this respect, a No-Reference Image Quality Assessment (NR-IQA) metric dedicated to uneven illumination is proposed in this paper. The key idea is to analyze the effect of contrast enhancement on the spatial distribution of the luminance component of the signal. The results obtained through extensive experiments, performed on a challenging dedicated database, have shown that the proposed metric significantly improves the state-of-the-art NR-IQA metrics when applied to this type of video content and distortion.

In this paper, we propose an effective Convolutional Autoencoder (AE) model for Sparse Representation (SR) in the Wavelet Domain for Classification (SRWC). The proposed approach involves an autoencoder with a sparse latent layer for learning sparse codes of wavelet features. The estimated sparse codes are used for assigning classes to test samples using a residual-based probabilistic criterion. Intensive experiments carried out on various datasets revealed that the proposed method yields better classification accuracy while exhibiting a significant reduction in the number of network parameters, compared to several recent deep learning-based methods.

This paper improves the conventional sparse representation based classification (SRC) method, through incorporating wavelet coefficients. For this reason, the proposed method is called Sparse Representation Wavelet based Classification (SRWC). In the present study, we fuse the image features described by the complementary information from the low sub-band of the wavelet coefficients and sparse representation to outperform the conventional SRC according to accuracy. This holds because the wavelets promote sparsity and provide structural information about the image, which increases the accuracy of classification. To validate the capabilities and underline the advantages of the novel SRWC, we conducted an extensive number of experiments using publicly available datasets and compared our results with contemporary methods.

Denoising is an essential application to improve image quality, especially in medical imaging. This paper introduces an example and patch-based learning method for reducing Gaussian noise and Poisson noise which often appear in medical imaging modalities using ionizing radiation. In the proposed method, denoising is performed by learning the regression model based on a set of the nearest neighbors of a given noisy patch, with the help of a given set of standard images. The method is evaluated and compared to several state-of-the-art denoising methods. The obtained results confirm its efficiency, especially for heavy noise.

In this paper, we propose a novel example-based method for denoising and super-resolution of medical images. The objective is to estimate a high-resolution image from a single noisy low-resolution image, with the help of a given database of high and low-resolution image patch pairs. Denoising and super-resolution in this paper is performed on each image patch. For each given input low-resolution patch, its high-resolution version is estimated based on finding a nonnegative sparse linear representation of the input patch over the low-resolution patches from the database, where the coefficients of the representation strongly depend on the similarity between the input patch and the sample patches in the database. The problem of finding the nonnegative sparse linear representation is modeled as a nonnegative quadratic programming problem. The proposed method is especially useful for the case of noise-corrupted and low-resolution image. Experimental results show that the proposed method outperforms other state-of-the-art super-resolution methods while effectively removing noise.

A new method for detecting and correcting color-timing mismatch in digital film is proposed. This method is based on a physical model which accounts for the absorption of light by the different layers of the film. The whole process consists of five sequential stages: shot change detection, registration, degraded zone detection, and unreliable motion detection and correction. One of the main issues in this paper is how to align the degraded and reference frames in the case of varying illumination conditions. A new transformation, based on the proposed physical model, is derived to make the data term of the optical flow framework robust to color change. The degraded regions are then detected and restored by using the computed flow field. The performance of the proposed method is evaluated through extensive tests on both simulated and real high definition films. The obtained results are very promising and confirm the efficiency of the proposed method.

Recently non-local means (NLM) has been known to be one of the most attractive denoising algorithms. It alters each pixel by a weighted average of pixels in the image. The weights express the level of similarity between two small patches defined for two involved pixels. There are many propositions to ameliorate the performance of this method. One of branches is to seek the whole image the most similar patches for a given one. In this paper, we investigate this approach and show that it is suitable for only highly textured images. Moreover, we show that combination of this approach and the original NLM yields better result for all image types.

Medical images are often corrupted by random noise, leading to undesirable visual quality. Thus, image denoising is one of the fundamental tasks required by medical imaging analysis. In this paper, we propose a novel learning method for the reduction of Gaussian noise of Computed Tomography (CT) image and Rician noise of Magnetic Resonance Imaging (MRI) image based on a given set of standard images and the Kernel Ridge Regression (KRR). Experimental results demonstrate the outperformance of the proposed technique over various other methods in terms of both objective and subjective evaluations.

A novel method for automatically detecting and restoring the color timing echo artifact which may result in the internegative/interpositive printing process is proposed. This degradation appears at shoot change, either on the top of the last frame of a shot or on the bottom of the subsequent frame of the next shot. In our work, the degraded region is firstly detected and then restored by using the histogram specification (HS) based method. To avoid discontinuities between the restored region and the remaining of the treated frame due to abrupt modification in a large region, the degraded zone is decomposed into several overlapped small blocks. Then a block matching algorithm is used and a block similarity metric based on ordering has also been proposed for motion estimation under luminance variation. The proposed method is tested on several color high definition (HD) films with highly promising results.

This paper presents a new perceptual image watermarking scheme based on the Laplacian pyramid (LP) decomposition and a visibility map constructed using a human visual contrast model. This map is computed of each level of the pyramid in order to determine the auspicious regions for embedding the watermark. The spread spectrum technique is used to embed the watermark in some levels of the LP. The watermarked image is then constructed from the Laplacian images. The algorithm performances are evaluated in terms of watermark invisibility, using an objective image quality perceptual measure, the structural similarity index measure (SSIM), and robustness to different attacks of Stirmark such as Jpeg compression, low-pass filtering, additive noise and cropping.

A new pyramidal JND (just noticeable difference) model is presented. The idea is to use this JND to determine the optimum strength for embedding the watermark providing an invisible and robust watermarking scheme. The image is first decomposed into a multiresolution representation using the pyramidal decomposition. Then, a perceptual model is proposed to compute the JND value for each pixel at each Laplacian level. This model takes into account three main characteristics of the human visual system (HVS), namely: contrast sensitivity, luminance adaptation and contrast masking. The performance of the proposed technique is evaluated in terms of transparency, using subjective and objective tests, and robustness to different common attacks.