The fundamental step to get a Statistical Shape Model (SSM) is to align all the training samples to the same spatial modality. In this paper, we propose a new 3D alignment method for organic training samples matching, whose modalities are orientable and surface figures could be recognized. It is a feature based alignment method which matches two models depending on the distribution of surface curvature. According to the affine transformation on 2D Gaussian map, the distances between the corresponding parts on surface could be minimized. We applied our proposed method on 5 cases left lung training samples alignment and 4 cases liver training samples alignment. The experiment results were performed on the left lung training samples and the liver training samples. The availability of proposed method was confirmed.
Recently, information technology (IT) has been introduced to social systems for various purposes. Especially, important information for our living has been brought by geographic information systems (GIS). Particularly, in the field of car navigation, frequent update of a map is requested for providing a driver with comprehensive and accurate geographic information having a driver's view. In this paper, we propose a simple and useful method of making a 3D geographical map by mapping onto a given 2D map the data measured in the street using video cameras and several sensors on a car. The function of the proposed method is to make a 3D map by fitting measured data to a 2D map, and not to make the 2D map. In the method, the data of road surfaces (road signs, directional arrows, etc.) and building textures are obtained simultaneously by the measurement system consisting of several video cameras, a distance pulse generator, and a GPS. In order to evaluate the proposed method, we performed experiments at Munich (Germany), and Fukuoka (Japan), and produced a highly realistic 3D map, which confirms the availability of the proposed method
In recent years, the death rate caused by lung cancer is increasing. To detect the lung cancer, multi detector-row computed tomography (MDCT) images are used in visual screening. Lung cancer can be easily detected by using the chest MDCT images, however, it has enormous images and burden to radiologists. Research and development of the computer aided diagnosis (CAD) system have been assisted the diagnosis. As one of the CAD technologies, temporal subtraction technique is possible to emphasize the changing interval on the CT images. It uses subtraction operation between previous and current CT images of the same patient. On the other hand, pattern recognition using image reconstruction by sparse coding method has attracted attention. This technique is mathematically modeling the information processing by the primary visual cortex of human. It is the technique for representing images by the linear combination of a small number of basis. In this paper, candidate nodules under 20[mm] were segmented from temporal subtraction images based on the 3D sparse coding technique.3D sparse coding is three dimensional expansion of the sparse coding. Also, we classified the final candidate nodules using support vector machine (SVM) method based on coefficient matrix which are obtained by the 3D sparse coding. We applied proposed method to 31 cases of chest MDCT images in which the number of nodules was more than one. We achieved experimental result with true positive rates (TPR) of 70.2[[%], and false positive rates (FP) of 34.7[/scan], respectively.
Recently, there are many traffic accidents in turning right at an intersection. They are mainly caused by a driver's oversight of pedestrians and motorcycles that are occluded by oncoming cars. Therefore a system is necessary to detect moving objects such as oncoming cars and pedestrians at an intersection, and warn a vehicle driver. This paper describes a technique for detecting moving objects in turning right at an intersection when vehicle is stopping. Moving objects are detected by Mixture of Gaussians (MoG). In addition, we distinguish cars from pedestrians using the difference of the area size and the aspect ratio of detected objects. The object which is classified as a pedestrian is tracked using Lucas-Kanade Tracker. If the detected cars and pedestrians overlap or a car completely obscures pedestrians, we perform the estimation of pedestrian's location by using the information on past frames. By doing this, it is possible to detect pedestrians that drivers are actually difficult to see. The performance of the proposed technique was examined employing car videos and satisfactory results were obtained.
Salient region of an image is usually detected by using contrast and boundary priors. Along with those cues the use of seam importance map has shown promising output previously. In this study, better result is found by further exploiting the seam-map using spatial distance and color information in combination with boundary prior. Color and seam maps are also down-weighted using average spatial distance to other regions. Moreover, passing the superpixelized version of the input image into seam and color map generation procedure has improved the output. Experimental results based on MSRA 1k dataset are presented with ten state of the art methods. F-beta measures are presented along with precision recall curves to better understand the outcome. The performance comparison with compared researches proofs superiority of the proposed method.
Object tracking is a challenging problem due to the presence of noise, occlusion, clutter and dynamic change in the scene other than the motion of the object of interest. A variety of tracking algorithms has been proposed and implemented to overcome the related difficulties, but there are still some problems need to be covered. In this paper, we present an approach for multiple objects tracking based on particle filter algorithm. We use the particle filter to predict the trajectory of the target. The problem of occlusion is predicted based on the likelihood measurement and estimated samples distance. The particle filter approximates a posterior probability density of the state using samples or particles. Each state is denoted as the hypothetical state of the tracked object and its weight which is predicted based on the system model. In this paper, the state is treated as a position, speed, size, scale and appearance of the object. The samples weight is considered as the likelihood of each particle which is measured based on the similarity between the colour feature of the target model and the objects. And finally, the mean state of the particles is treated as the estimated state of the object. The experiments are performed to confirm the effectiveness of the method to track multiple objects.
'/%3e%3cuse xlink:href='%23a' transform='translate(0 -400)'/%3e%3cuse xlink:href='%23a' transform='translate(0 -600)'/%3e%3cuse xlink:href='%23a' transform='translate(0 -800)'/%3e%3cuse xlink:href='%23a' transform='translate(0 -1000)'/%3e%3cuse xlink:href='%23a' transform='translate(0 -1200)'/%3e%3cpath d='M0 0h1400v800H0z' style='fill:%23010066'/%3e%3cpath d='M576 100c-166.146 0-301 134.406-301 300s134.854 300 301 300c60.027 0 115.955-17.564 162.927-47.783-27.353 9.44-56.71 14.602-87.271 14.602-147.327 0-266.897-119.172-266.897-266.01 0-146.837 119.57-266.01 266.897-266.01 32.558 0 63.746 5.815 92.602 16.468C696.217 118.91 638.305 100 576 100z' style='fill:%23fc0'/%3e%3cpath d='m914.286 471.429-99.538-53.25 29.43 108.982-66.575-91.165-20.77 110.96-20.428-111.024-66.857 90.96L699.314 418l-99.701 52.943 74.065-85.192-112.8 4.441 103.694-44.62-103.555-44.94L673.8 305.42 600 220l99.538 53.25-29.43-108.982 66.575 91.165 20.77-110.96 20.428 111.023 66.857-90.959L814.97 273.43l99.702-52.944-74.065 85.193 112.8-4.441-103.694 44.62 103.555 44.94-112.785-4.79z' style='fill:%23fc0' transform='matrix(1.2738 0 0 1.2423 -89.443 -29.478)'/%3e%3c/svg%3e)