Sorry, you need to enable JavaScript to visit this website.

Fisheye camera is widely used in various applications because of its wider field-of-view. However, high distortion of images captured by fisheye cameras make it difficult for certain tasks which are based on traditional epipolar geometry (using epipolar lines) and stereo correspondence, such as depth map estimation. While most of existing depth map estimation methods use perspective-projection-based camera model, considering fisheye camera for depth map estimation will be beneficial because of its wider FOV.

We presents a no-reference (NR) image sharpness metric based on a visual sensitivity model. We propose that MaxPol convolution kernels are close approximation to this model and capable of extracting meaningful features for image sharpness assessment. Equipped by these kernels, we develop an efficient pipeline to evaluate the out-of-focus level of input images by decomposing the first and third order image differentials. The associated kernels are regulated in higher cutoff frequencies to balance out the information loss and noise sensitivity.

This paper presents a novel algorithm for fast and effective vanishing point detection. Once line segments in an input image are detected by LSD algorithm, the proposed method filters out outlier line segments. The remaining line segments are then over-clustered, and each cluster is assigned to 5 different types. According to the assigned type, each cluster is re-merged by applying different criteria, and the re-merged clusters generate hypotheses for vanishing points.

This paper presents a novel Time-of-Flight (ToF) depth recovery algorithm minimizing a new quadratic energy function utilizing depth and infrared data. The proposed energy function consists of a filtering term and a reconstruction term to remove noise and fill holes simultaneously in a depth image. In the filtering term, a new multilateral weight is introduced by fully using available spatial, depth, and infra¬red information.

This paper proposes a statistical method to enhance image quality in order to increase the click-through rate (CTR) of product images. We build a joint probability model of global image features for photos of different product categories. The images are modified in terms of brightness, contrast, and sharpness in order to increase the expected CTR. The effectiveness of the method is evaluated using a perceptual user study, comparing it to histogram equalization methods, and by conducting an A/B test over one week on the e-commerce site Rakuten Ichiba.

The orthogonal matching pursuit (OMP) is an important sparse approximation algorithm to recover sparse signals from compressed measurements. However, most MP algorithms are based on the mean square error(MSE) to minimize the recovery error, which is suboptimal when there are outliers.

This paper proposes a stacked conditional Generative Adversarial Network-based method for Near InfraRed (NIR) imagery colorization. We propose a variant architecture of Generative Adversarial Network (GAN) that uses multiple loss functions over a conditional probabilistic generative model. We show that this new architecture/loss-function yields better generalization and representation of the generated colored IR images. The proposed approach is evaluated on a large test dataset and compared to recent state of the art methods using standard metrics.

Coded exposure photography (CEP), originally proposed by Raskar et al., has been known as one of the promising techniques for motion deblurring. In this area, much efforts have been made for designing a fluttered shutter sequence to shape the spectrum of a uniformly motion-blurred image into an invertible one. Since the duty cycle of the fluttered shutters proposed thus far is generally low, the number of photons entering into an image sensor is reduced, which leads to a large Poisson noise in a low lighting condition.

• A connected-tube model based on a Marked Point Process (MPP) for strip feature extraction in images is proposed
• A mixed MPP model can be formed by combing the proposed model with other geometric models
• The proposed model can be applied to complex detection tasks, such as short and long fiber detection in material images, road and roof detection in satellite images.