A Contour Detection Method Based on Image Saliency

Person detection is an important prerequisite of social interaction, but is not well understood. Following suggestions that people in the visual field can capture a viewer's attention, this study examines the role of the face and the body for person detection in natural scenes. We observed that viewers tend first to look at the center of a scene, and only then to fixate on a person. When a person's face was rendered invisible in scenes, bodies were detected as quickly as faces without bodies, indicating that both are equally useful for person detection. Detection was optimized when face and body could be seen, but observers preferentially fixated faces, reinforcing the notion of a prominent role for the face in social perception. These findings have implications for claims of attention capture by faces in that they demonstrate a mediating influence of body cues and general scanning principles in natural scenes.

Text detection in natural scenes is fundamental for text image analysis. In this paper, we propose a context-based approach for robust and fast text detection. Our main contribution is that we introduce a new concept of key region, which is described with context according to stroke properties, appearance consistency and specific spatial distribution of text line. With such context descriptors, we adopt SVM to learn a context-based classifier to find key regions in candidate regions. Therein, candidate regions are connected components generated by local binarization algorithm in the areas, which are detected by an offline learned text patch detector. Experimental results on two benchmark datasets demonstrate that our approach has achieved competitive performances compared with the state-of-the-art algorithms including the stroke width transform (SWT) [1] and the hybrid approach based on CRFs [2] with speedup rates of about 1.7x~4.4x.

This paper investigates two fundamental problems in computer vision: contour detection and image segmentation. We present state-of-the-art algorithms for both of these tasks. Our contour detector combines multiple local cues into a globalization framework based on spectral clustering. Our segmentation algorithm consists of generic machinery for transforming the output of any contour detector into a hierarchical region tree. In this manner, we reduce the problem of image segmentation to that of contour detection. Extensive experimental evaluation demonstrates that both our contour detection and segmentation methods significantly outperform competing algorithms. The automatically generated hierarchical segmentations can be interactively refined by user-specified annotations. Computation at multiple image resolutions provides a means of coupling our system to recognition applications.

An approach to symbolic contour extraction is described that consists of three stages: enhancement, detection, and extraction of contours and corners. Contours and corners are enhanced by models of monkey cortical complex and endstopped cells. Detection of corners and local contour maxima is performed by selection of local maxima in both contour and corner enhanced images. These maxima form the anchor points of a greedy contour following algorithm that extracts the contours. This algorithm is based on the idea of spatially linking neurons along a contour that fire in synchrony to indicate an extracted contour. The extracted contours and detected corners represent the symbolic representation of the image. The advantage of the proposed model over other models is that the same low constant thresholds for corner and local contour maxima detection are used for different images. Closed contours are guaranteed by the contour following algorithm to yield a fully symbolic representation which is more suitable for reasoning and recognition. In this respect our methodology is unique, and clearly different from the standard (edge) contour detection methods. The results of the extracted contours (when displayed as being detected) show similar or better results compared to the SUSAN and Canny-CSS detectors.

The traditional method to extract target contour from aerial target image is changing the aerial image into a gray level image with multiple thresholds or binary image with single threshold. From the edge of target, contour can be extracted according to the changed value. The traditional method is useful only when contrast between target and background is in the proper degree. Snakes are curves defined within an image domain that can move under the influence of internal force coming from within the curve itself and external forces are defined so that the snake will conform to an object boundary or other desired features within an image. Snakes have been proved an effective method and widely used in image processing and computer vision. Snakes synthesize parametric curves within an image domain and allow them to move toward desired edges. Particular advantages of the GVF(Gradient Vector Flow) snakes over a traditional snakes are its insensitivity to initialization and its ability to move into boundary concavities. Its initializations can be inside, outside, or across the object’s boundary. The GVF snake does not need prior knowledge about whether to shrink or expand toward the boundary. This increased capture range is achieved through a diffusion process that does not blur the edges of themselves. Affected by the light from different incident angle, the brightness of aerial target surface changed greatly in a complicate mode. So the GVF snakes is not fast, accurate and effective all the time for this kind of images. A new contour extracting method, GVF Snakes Combined with wavelet multi-resolution Analysis is proposed in this paper. In this algorithm, bubble wavelet is used iteratively to do the multi resolution analysis in the order of degressive scale before GVF Snakes is used every time to extract accurate contour of target. After accurate contour is extracted, polygon approximation is used to extract characteristics to realize the recognition of aerial target. The process is in the following: Step 1: use bubble wavelet filter to cut big part of the noises, weakening false edges. Step 2: initialize active contour and control the contour’s move according to GVF to get a new contour. Step 3: decrease the scale of filter, and use the new contour as the initial contour and control the contour’s move to get new contour again. Step 4: repeat step 3 till the set scale is reached. The last new contour is the final contour. Step 5: find the center determine an axis by calculate distance between every point on the final contour to the center. Step 6: adjust the distance threshold and combine the points until the contour is changed into a polygon with fixed angle number which is best fit the target recognition demand. Step 7: use the polygon to match the target plate to recognize target. Applied the new algorithm to aerial target images of a helicopter and a F22 battleplan, the contour extraction and polygon approximation results show that targets can be matched and recognized successfully. This paper mainly focuses on contour extraction and polygon approximation in the recognition area.

As an important core in the intelligent traffic management system, the technology and application of license plate recognition have become research focus. Detecting the accurate location of a license plate from a vehicle image is considered to be the most crucial step of license plate recognition, which greatly affects the recognition rate and speed of the whole system. Nevertheless, due to the low accuracy of license plate detection in natural scenes, further investigations are still needed in this field in order to make the detection process very efficient. In this paper, We have studied and implemented a convolutional neural network license plate detection algorithm based on transfer learning. According to the invention, new energy license plates and ordinary license plates are adopted as the research objects. The text detection model AdvancedEAST is trained on the license plate images through the transfer learning method, and experiments are carried out on the self-built license plate dataset. The experimental results show that the algorithm can better adapt to light complexity, low resolution, target interference, license plate tilt and other complex conditions. The license plate positioning algorithm has high accuracy in natural scenes, and it is superior to the traditional license plate detection methods.

Efficient local and global contour detection based on superpixels

Contour detection via stacking random forest learning

We investigate the contour detection task in complex natural images. We propose a novel contour detection algorithm which jointly tracks at two scales small pieces of edges called edgelets. This multiscale edgelet structure naturally embeds semi-local information and is the basic element of the proposed recursive Bayesian modeling. Prior and transition distributions are learned offline using a shape database. Likelihood functions are learned online, thus are adaptive to an image, and integrate color and gradient information via local, textural, oriented, and profile gradient-based features. The underlying model is estimated using a sequential Monte Carlo approach, and the final soft contour detection map is retrieved from the approximated trajectory distribution. We also propose to extend the model to the interactive cut-out task. Experiments conducted on the Berkeley Segmentation data sets show that the proposed MultiScale Particle Filter Contour Detector method performs well compared to competing state-of-the-art methods.

Recent research on neocognitron-like neural feed-forward architectures, which have formerly been successfully applied to the recognition of artificial stimuli such as paperclip objects, now also opens up application to more natural stimuli. Such networks exhibit high-recognition performance with respect to translation, rotation, scaling and cluttered surroundings. In this contribution, we introduce a new type of hierarchical model, which is trained using a non-negative matrix factorization algorithm. In contrast to previous work, our approach cannot only classify objects but is also capable of rapid object detection in natural scenes. Thus, the time-consuming and conceptually unsatisfying split-up into a localization stage (e.g., using segmentation) and a subsequent classification can be avoided. The network consists of alternating layers of simple and complex cell planes and incorporates nonlinear processing schemes that have been proposed in recent literature. Learning of receptive field profiles for the lower layers of the network takes place by unsupervised learning whereas a final classification layer is trained supervised. This final layer is then utilized for detection. We test the classification performance of the network on images of natural objects which are systematically distorted. To test the ability to detect objects, cluttered natural background is used.

Text in images provides rich and accurate high-level semantic information, which is of great significance for many potential applications such as scene understanding, image and video retrieval. As a necessary means of understanding scene text, text detection has been widely concerned in the field of computer vision in recent years. However, due to the diversity of text scale and the uncertainty of image quality, text detection in natural scenes is still a challenging problem. Inspired by the successful application of deep neural network in target detection task, this paper proposes a text detection method based on multi-level feature extraction combined with deep learning technology. The multi-level feature extraction designed by this method can capture the details of the text, which is conducive to enhance the robustness of the algorithm to small-scale text and low-resolution images. In addition, the method is end-to-end, and can output text detection results directly. This paper evaluates the method on several standard datasets, and the experimental results prove the effectiveness of the proposed method.

With the worldwide spread of COVID-19, people's life safety has been greatly threatened. So, we consider using YOLOv3-tiny algorithm to detect mask wearing. Since there are few detection models for correctly wearing masks, we decided to use three classifications to detect correctly wearing masks, incorrectly wearing masks, and not wearing masks. Besides, in order to enhance the performance of our model in small object detection, we propose the k-means++ algorithm to make the size of the initial anchor boxes closer to the actual size of the object, and add a YOLO detection layer to effectively improve the accuracy of a small object. The results show that the mAP@50 values of our model are 4.68% higher than YOLOv3-tiny algorithm. Our model has significantly improved the detection ability of crowd scenes, and mask detection is more accurate and robust, which has good application value for mask detection in natural scenes.

Multi-scale object detection in natural scenes is still challenging. To enhance the multi-scale perception capability, some algorithms combine the lower-level and higher-level information via multi-scale feature fusion strategies. However, the inherent spatial properties among instances and relations between foreground and background are ignored. In addition, the human-defined “center-based” regression quality evaluation strategy, predicting a high-to-low score based on a linear relationship with the distance to the center of ground-truth box, is not robust to scale-variant objects. In this work, we propose a Depth-Guided Progressive Network (DGPNet) for multi-scale object detection. Specifically, besides the prediction of classification and localization, the depth is estimated and used to guide the image features in a weighted manner to obtain a better spatial representation. Therefore, depth estimation and 2D object detection are simultaneously learned via a unified network, where the depth features are merged as auxiliary information into the detection branch to enhance the discrimination among multi-scale objects. Moreover, to overcome the difficulty of empirically fitting the localization quality function, high-quality predicted boxes on scale-variant objects are more adaptively obtained by an IoU-aware progressive sampling strategy. We divide the sampling process into two stages, i.e., “statistical-aware” and “IoU-aware”. The former selects thresholds for positive samples based on statistical characteristics of multi-scale instances, and the latter further selects high-quality samples by IoU on the basis of the former. Therefore, the final ranking scores better reflect the quality of localization. Experiments verify that our method outperforms state-of-the-art methods on the KINS and Cityscapes dataset.

A systematic comparison between visual cues for boundary detection

In this paper, we consider water surface object detection in natural scenes. Generally, background subtraction and image segmentation are the classical object detection methods. The former is highly susceptible to variable scenes, so its accuracy will be greatly reduced when detecting water surface objects due to the changing of the sunlight and waves. The latter is more sensitive to the selection of object features, which will lead to poor generalization as a result, so it cannot be applied widely. Consequently, methods based on deep learning have recently been proposed. The River Chief System has been implemented in China recently, and one of the important requirements is to detect and deal with the water surface floats in a timely fashion. In response to this case, we propose a real-time water surface object detection method in this paper which is based on the Faster R-CNN. The proposed network model includes two modules and integrates low-level features with high-level features to improve detection accuracy. Moreover, we propose to set the different scales and aspect ratios of anchors by analyzing the distribution of object scales in our dataset, so our method has good robustness and high detection accuracy for multi-scale objects in complex natural scenes. We utilized the proposed method to detect the floats on the water surface via a three-day video surveillance stream of the North Canal in Beijing, and validated its performance. The experiments show that the mean average precision (MAP) of the proposed method was 83.7%, and the detection speed was 13 frames per second. Therefore, our method can be applied in complex natural scenes and mostly meets the requirements of accuracy and speed of water surface object detection online.