Detection and segmentation of moving objects in complex scenes
نویسندگان
چکیده
Detecting and segmenting moving objects in dynamic scenes is a hard but essential task in a large number of applications such as surveillance. Most existing methods only give good results in the case of persistent or slowly changing background, or if both the objects and the background can be characterized by simple parametric motions. This paper aims at detecting and segmenting foreground moving objects in the absence of such constraints. The sequences we consider have highly dynamic backgrounds, illumination changes and low contrasts, and can have been shot by a moving camera. Three main steps compose the proposed method. First, moving points are selected within a sub-grid of image pixels. A descriptor is associated to each of these points. Clusters of points are then formed using a variable bandwidth mean shift with automatic bandwidth selection. Finally, segmentation of the object associated to a given cluster is performed using Graph cuts. Experiments and comparison to other motion detection methods on challenging sequences show the performance of the proposed method and its utility for video analysis in complex scenes. Key-words: motion detection, segmentation, mean shift clustering, graph cuts Détection et segmentation d’objets en mouvement dans des scènes complexes Résumé : De nombreuses applications en vision par ordinateur et en surveillance nécessitent la détection et la segmentation des objets en mouvement. La plupart des méthodes existantes ne donnent de bons résultats que pour des fonds statiques ou peu changeants, ou si le fond et les objets sont rigides et ont un mouvement affine 2D. Le but de ce papier est de directement détecter les objets en mouvement dans des séquences complexes n’ayant pas ces caractéristiques. Les vidéos considérées ici ont un fond dynamique, avec de forts changements d’illumination et de faibles contrastes, et peuvent avoir été prises par une caméra en mouvement. La méthode proposée se divise en trois étapes principales. Tout d’abord un ensemble de points en mouvement est sélectionné parmi une grille de pixels uniformément répartis sur toute l’image. Tous ces points sont associés à un descripteur. La deuxième étape consiste à former des groupes de ces points représentant chacun un objet en mouvement. Ces partitions sont obtenues par un algorithme mean shift à noyau variable avec une sélection automatique de la taille du noyau. Enfin, à partir de ces groupes de points, la segmentation des objets est donnée en minimisant une énergie par coupure de graphe. Des résultats et comparaisons avec d’autres méthodes de segmentation de mouvement montrent l’efficacité de la méthode proposée. Mots-clés : détection de mouvement, segmentation, partitionnement mean shift, coupure de graphe Detection and segmentation of moving objects in complex scenes 3
منابع مشابه
Statistical Background Modeling Based on Velocity and Orientation of Moving Objects
Background modeling is an important step in moving object detection and tracking. In this paper, we propose a new statistical approach in which, a sequence of frames are selected according to velocity and direction of some moving objects and then an initial background is modeled, based on the detection of gray pixel's value changes. To have used this sequence of frames, no estimator or distribu...
متن کاملMoving Objects Tracking Using Statistical Models
Object detection plays an important role in successfulness of a wide range of applications that involve images as input data. In this paper we have presented a new approach for background modeling by nonconsecutive frames differencing. Direction and velocity of moving objects have been extracted in order to get an appropriate sequence of frames to perform frame subtraction. Stationary parts of ...
متن کاملMoving Objects Tracking Using Statistical Models
Object detection plays an important role in successfulness of a wide range of applications that involve images as input data. In this paper we have presented a new approach for background modeling by nonconsecutive frames differencing. Direction and velocity of moving objects have been extracted in order to get an appropriate sequence of frames to perform frame subtraction. Stationary parts of ...
متن کاملA Probabilistic Framework Based on KDE-GMM Hybrid Model for Moving Object Segmentation in Dynamic Scenes
In real scenes, dynamic background and moving cast shadow always make accurate moving object detection difficult. In this paper, a probabilistic framework for moving object segmentation in dynamic scenes is proposed. Under this framework, we deal with foreground detection and shadow removal simultaneously by constructing probability density functions (PDFs) of moving objects and non-moving obje...
متن کاملPlant Classification in Images of Natural Scenes Using Segmentations Fusion
This paper presents a novel approach to automatic classifying and identifying of tree leaves using image segmentation fusion. With the development of mobile devices and remote access, automatic plant identification in images taken in natural scenes has received much attention. Image segmentation plays a key role in most plant identification methods, especially in complex background images. Wher...
متن کاملA multi-scale convolutional neural network for automatic cloud and cloud shadow detection from Gaofen-1 images
The reconstruction of the information contaminated by cloud and cloud shadow is an important step in pre-processing of high-resolution satellite images. The cloud and cloud shadow automatic segmentation could be the first step in the process of reconstructing the information contaminated by cloud and cloud shadow. This stage is a remarkable challenge due to the relatively inefficient performanc...
متن کاملذخیره در منابع من
با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید
برای دانلود متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید
ثبت ناماگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید
ورودعنوان ژورنال:
- Computer Vision and Image Understanding
دوره 113 شماره
صفحات -
تاریخ انتشار 2009