Shota Nakashima

Single image dehazing through improved atmospheric light estimation

Image contrast enhancement for outdoor vision is important for smart car auxiliary transport systems. The video frames captured in poor weather conditions are often characterized by poor visibility. Most image dehazing algorithms consider to use a hard threshold assumptions or user input to estimate atmospheric light. However, the brightest pixels sometimes are objects such as car lights or streetlights, especially for smart car auxiliary transport systems. Simply using a hard threshold may cause a wrong estimation. In this paper, we propose a single optimized image dehazing method that estimates atmospheric light efficiently and removes haze through the estimation of a semi-globally adaptive filter. The enhanced images are characterized with little noise and good exposure in dark regions. The textures and edges of the processed images are also enhanced significantly.

Underwater Image Super-Resolution by Descattering and Fusion

Underwater images are degraded due to scatters and absorption, resulting in low contrast and color distortion. In this paper, a novel self-similarity-based method for descattering and super resolution (SR) of underwater images is proposed. The traditional approach of preprocessing the image using a descattering algorithm, followed by application of an SR method, has the limitation that most of the high-frequency information is lost during descattering. Consequently, we propose a novel high turbidity underwater image SR algorithm. We first obtain a high resolution (HR) image of scattered and descattered images by using a self-similarity-based SR algorithm. Next, we apply a convex fusion rule for recovering the final HR image. The super-resolved images have a reasonable noise level after descattering and demonstrate visually more pleasing results than conventional approaches. Furthermore, numerical metrics demonstrate that the proposed algorithm shows a consistent improvement and that edges are significantly enhanced.

An Improved Method for CT/MRI Image Fusion on Bandelets Transform Domain

People living in the information age, are more and more attention to their own lives. It is also said, social life is more important in present and future. The social life contains three fields. In this paper, we will propose a new method for adjunctive therapy in social life. Recent years, as the bandelets transform has some benefits, many scholars are interested in this field. They proposed many methods to solve different problems in different fields. In this paper, we propose a new maximum local energy method to calculate the low coefficients of images. And then adopt the sum modified laplacian method to select the high coefficients of images. Later, we compare the results with wedgelets transform. In our experiments, we take wedgelets transform, bandelets transform, and LE-wedgelets transform for comparing the results. Beside the human vision, we also compare the results by quantitative analysis. The numerical experiments state clearly that the maximum local energy is an effect way for image fusion, which can get well performance in visual effect and quantitative analysis. During 100 clinic CT/MR fusion experiments in practice, compare with previous methods, the PSNR of our method is improved respectively 5.836, 5.337, 0.035.

Slit Type One-Dimensional Brightness Distribution Sensor

As aging society problem goes serious, systems to confirm to safety of elders in daily life are expected. In this paper, a sensor, which detects person localization without privacy offending, applying brightness distribution is proposed. In the proposed design, the sensor is constructed with a line sensor and slit to obtain one-dimensional brightness distribution. The proposed sensor is able to obtain one-dimensional brightness distribution that is approximately equal to integration value of each vertical pixel line of two-dimensional image. By employing this method, the information of the targets position and movement status can be obtained without using two-dimensional detail texture image. In this paper, the effectiveness of the proposed method is confirmed by experiments.

Slit-type one-dimensional brightness distribution sensor for object tracking

This paper presents a novel computational imaging method for object tracking. The proposed slit-type one-dimensional brightness distribution sensor (Obrid-Sensor) is designed to be applied to detect falls, localize, and track subjects without invading privacy. According to the vertical and horizontal one-dimensional brightness distributions acquired by the proposed sensors, the subjects motion is monitored without images or videos. The effectiveness of the proposed method to detect falls, localize, and track is verified by experiment. The motion detection and location can be implemented without requiring images or videos, which may invade privacy. The detection range of the proposed privacy-preserving sensor is also confirmed. Finally, the proposed sensor has a simpler structure than the traditional Obrid-Sensor, making it attractive owing to its simple design, facile application, and low cost.

Detection range fitting of slit type Obrid-Sensor

As aging society problem goes severe, systems to confirm to safety of elders in daily life are expected to relieve burdensome safety confirmation tasks of care workers. In this paper, a sensor, which detects person localization without privacy offending, applying Obrid-Sensor is proposed. In the proposed design, the Obrid-Sensor is constructed with a line sensor and a slit to obtain one-dimensional brightness distribution. The proposed sensor is able to obtain one-dimensional brightness distribution that is approximately equal to integration value of each vertical pixel line of two-dimensional image. Meanwhile, the novel slit type Obrid-Sensor, which was constructed without Rod lens, is studied in this research. By employing the proposed sensor, the information of a subjects position and motion can be obtained without using two-dimensional texture image. The effectiveness of the proposed method is confirmed by experiments.

Multimodal Medical Image Fusion in Extended Contourlet Transform Domain

As a novel of multi-resolution analysis tool, the modified sharp frequency localized contourlet transforms (MSFLCT) provides flexible multiresolution, anisotropy, and directional expansion for medical images. In this paper, we proposed a new fusion rule for multimodal medical images based on MSFLCT. The multimodal medical images are decomposed by MSFLCT. For the high-pass subband, the weighted sum modified laplacian (WSML) method is used for choose the high frequency coefficients. For the lowpass subband, the maximum local energy (MLE) method is combined with “region” idea for low frequency coefficient selection. The final fusion image is obtained by applying inverse MSFLCT to fused lowpass and highpass subbands. Abundant experiments have been made on groups of multimodality datasets, both human visual and quantitative analysis show that the new strategy for attaining image fusion with satisfactory performance.

A New Type of Using Morphology Methods to Detect Blood Cancer Cells

In order to resolve the problem of recognizing blood cancer cells accurately and effectively, an identifying and classifying algorithm was proposed using grey level and color space. After image processing, blood cells images were gained by using denoising, smoothness, image erosion and so on. After that, we use granularity analysis method and morphology to recognize the blood cells. And then, calculate four characterizes of each cell, which is, area, roundness, rectangle factor and elongation, to analysis the cells. Moreover, we also applied the chromatic features to recognize the blood cancer cells. The algorithm was testified in many clinical collected cases of blood cells images. The results proved that the algorithm was valid and efficient in recognizing blood cancer cells and had relatively high accurate rates on identification and classification.

A Monitoring System to Prevent Forgetting to Take Medicines on Time Using Funnel-Shaped Sensors

To prevent people to forget to take medicines, we propose a new monitoring system to prevent forgetting to take medicines on time. In this system, we develop a funnel-shaped sensor consisting of a holder with an LED below the lid of the holder and a photo detector at the bottom to detect medicines. When medicines are dropped into the sensor, they fall down to the bottom of the sensor and the photo detector receives only part light of the LED. According to the value of the output of the photo detector, it can be easily to judge medicines with a proper threshold value. Highly accurate monitoring system to prevent forgetting to take medicines can be produced by using this sensor. If medicines are not taken at time to take medicines, the system announces to the user.

A Proposal of High-Performance Method for Distance Measuring Sensor United with PSD

In this study, we propose a sensor system to improve the performance of distance measurement using multiple paralleled position sensitive detector (PSD) sensors. The PSD sensor uses a light emitted diode (LED) to illuminate an object and measures the distance to the object by the position on the PSD of the gathered light from the object. The proposed sensor system uses two symmetrically paralleled PSD sensors to increase the measuring distance extent and accuracy. Experimental results showed that the proposed PSD sensor system has an enlarged measuring extent, and has a measuring accuracy better than or equal to the conventional one.