Salient target detection method under sea surface environment based on multi-scale phase spectrum

Deep (machine) learning in recent years has significantly increased the predictive modeling strengths for applications in the areas of computer vision, speech recognition, and natural language processing. Being heavily data-driven, conventional general-purpose deep learning methods mainly rely on training data to learn task-specific data representations. In this thesis, we focus on leveraging novel learning algorithms in deep learning to tackle visual recognition challenges at the three key levels -- namely pixel, object, and image levels. Throughout the thesis, we demonstrate how each and every of our proposed methods can address a specific visual recognition problem. Firstly, we apply deep learning to improve the saliency detection performance of deep networks. A coarse saliency map of the input image's salient object is initially predicted by a convolutional-deconvolutional network (or equivalently a fully convolutional network). Following that, the coarse saliency map is sequentially refined by a separate network called Recurrent Attentional Convolutional-Deconvolutional Network (RACDNN) that shares its network weights across its sequential steps. At every sequential (temporal) step, the same RACDNN is applied to a newly selected/attended image sub-region and it performs saliency map refinement for that particular sub-region. Despite having many learnable network parameters and no access to additional training data, RADCNN can be trained well and it outperforms the "single-step" (coarse saliency detection) baseline and state-of-the-art saliency detection methods by large margins, on several saliency detection benchmark datasets. Secondly, we introduce auto-encoder weight transfer networks (AE-WTN+), a deep model for scaling up object detection. Conventional WTN transfers object class knowledge from the classification weights of a pretrained large-scale image classification network, to the classification weights of an object detection network. However, they suffer from under-fitting and are not able to preserve the rich object class information of novel classes. In this work, we propose a new WTN method -- AE-WTN+ which makes use of normalization techniques to ease network optimization. Inputs and intermediate features are normalized to improve trainability for reduced under-fitting. More importantly, a novel auto-encoding (reconstruction) loss is proposed to encourage information preservation of all classes in the weight predictions of AE-WTN+. This enables the predicted weights of AE-WTN+ to maintain the class discriminability and neighbourhood relationships contained in the pre-trained large-scale classification weights. Experiments on large-scale detection datasets validate the effectiveness of AE-WTN+ on both seen and novel classes. Thirdly, we improve the efficiency of cross-layer network connections for very deep convolutional network in the context of image classification. This gives rise to a new convolutional network architecture called Deluge Networks (DelugeNets) which pass the output (feature activations) of any preceding composite network layer as direct inputs to all succeeding composite layers. For the incoming feature activations of preceding composite layers, the succeeding composite layers first apply rather lightweight cross-layer depthwise convolutional weights on them for feature aggregation. The heavy lifting is carried out by the last convolutional weight layer in each preceding composite layer that performs cross-channel convolutions. The proposed technique enables unimpeded information flow among many layers in very deep DelugeNets with great efficiency. Experiments on small-scale and large-scale datasets show that DelugeNets perform competitively to state-of-the-art deep convolutional networks at lower computational and parameter costs. Fourthly, we propose Stochastic Downsampling (SDPoint), a method to overcome a major limitation suffered by conventional deep convolutional networks -- their computational costs during inference are fixed and cannot be changed. SDPoint downsamples the feature maps at a randomly selected point (layer index) in the network hierarchy during training. The many downsampling configurations known as SDPoint instances each entail a unique downsampling point and downsampling ratio. The same convolutional network weights are shared across the different SDPoint instances that downsample feature maps differently. During inference, the computational cost of a network trained with SDPoint can be instantaneously adjusted to fit a given computational budget, by manually selecting an appropriate SDPoint instance. Additionally, sharing weights between different feature map scales provides significant regularization benefits, making the networks trained with SDPoint less sensitive to scales and more competent. In image classification experiments, SDPoint demonstrates significant cost-accuracy and computational advantages, over independently-trained (non-weight-shared) baselines with various inference costs.

This study presents an efficient saliency model mainly aiming at content-based applications such as salient object segmentation. The input colour image is first pre-segmented into a set of regions using the mean shift algorithm. A set of Gaussian models are estimated on the basis of segmented regions, and then for each pixel, a set of normalised colour likelihood measures to different Gaussian models are calculated. The colour saliency measure and spatial saliency measure of each Gaussian model are evaluated based on its colour distinctiveness and the spatial distribution, respectively. Finally, the pixel-wise colour saliency map and spatial saliency map are generated by summing the colour and spatial saliency measures of Gaussian models weighted by the normalised colour likelihood measures, and they are further combined to obtain the final saliency map. Experimental results on a dataset with 1000 images and ground truths demonstrate the better saliency detection performance of our saliency model.

Frequency-tuned saliency detection analyzes image saliency from the perspective of frequency domain and fully combines image segmentation method, which outputs well- defined boundaries of salient objects. However, the method ignores spatial relationships across image parts. This paper proposes an improved saliency detection method on the basis of the frequency-tuned method. In this method, we first segment the input image into regions and then analyze the image from the frequency domain. After that, we preprocess it using Gaussian filter to eliminate noise and coding artifacts. For each region, we can get saliency map in region-level based on region contrast. Finally, salient regions are selected by winner-take-all (WTA) neural network and Inhibition of Return(IOR) mechanism. The proposed salient region detection algorithm combines the virtues of frequency-tuning and region contrast. The experimental results show the feasibility and validity of this algorithm. Keywords-Image retrieval; Semantic gap; Saliency detection; Region-level saliency map; Visual attention

Biologically inspired visual object tracking has been very popular in the last decades. Several methods have exploited some visual cognitive mechanisms to outperform trackers purely based on machine vision in accuracy and speed. Although they have shown outstanding performance on recent benchmarks, they are still far from achieving the performance of a human in tracking. Visual attention is an important mechanism, which can be applied by saliency detection models. Saliency detection models can be determined in the spatial domain or the frequency domain. Spatial domain based models mostly have complex computations and are time-consuming. Thus, we have improved and recruited Frequency Divisive Normalization (FDN) model as a saliency detection method to propose a bio-inspired object tracking algorithm. This algorithm has been developed based on Fast Fourier Transform (FFT) and uses spectral features in order to achieve a high-speed tracker. It also can be launched on normal hardware. FDN model is based on Feature Integration Theory (FIT) and operates all the processes of saliency detection in the frequency domain, including decomposing the image to the features, lateral inhibition interaction, combining all features, and computing the saliency map. In addition to, extracting color, intensity and orientation it uses spatial frequency features, which is rarely used by other trackers. Extensive evaluations on large-scale benchmark datasets show that the proposed method has good performance with respect to its runtime.

Despite the advantage of exploiting interimage information by performing joint processing of images for co-saliency, co-segmentation, or co-localization, it introduces a few drawbacks: 1) its necessity in scenarios where the joint processing might not perform better than individual image processing; 2) increased complexity over individual image processing; and 3) complex parameter tuning. In this paper, we propose a simple cosaliency estimation method where we fuse saliency maps of different images using the dense correspondence technique. More important, the co-saliency estimation is guided by our proposed quality measurement that helps decide whether the saliency fusion really improves the quality of the saliency map or not. Our basic idea for developing the quality metric is that a high-quality saliency map should have well-separated foreground and background, as well as a concentrated foreground like ground-truths. Extensive experiments on several benchmark datasets including the large-scale dataset, ImageNet, for the applications of foreground co-segmentation and co-localization show that our proposed framework is able to achieve very competitive results.

Saliency detection is a technique to analyze image surroundings to extract relevant regions from the background. In this paper, we propose a simple and effective saliency detection method based on image sparse representation and color features combination. First, the input image is segmented into non-overlapping super-pixels, so as to perform the saliency detection at the region level to reduce computational complexity. Then, a background optimization selection scheme is used to construct an appropriate background template. Based on this, a primary saliency map is obtained by using image sparse representation. Next, through the linear combination of color coefficients we generate an improved saliency map with more prominent salient regions. Finally, the two saliency maps are integrated within Bayesian framework to obtain the final saliency map. Experimental results show that the proposed method has desirable detection performance in terms of detection accuracy and running time.

Saliency and co-saliency detection aim to distinguish conspicuous foreground objects from single and multiple images, thus are essential in many multimedia and vision applications. To achieve balanced efficiency and accuracy, most recent successful saliency detectors are based on superpixels. However, saliency detection with single-scale superpixel segmentation may fail in capturing intrinsic salient objects of complex natural scenes with small-scale high-contrast backgrounds. To tackle this problem, we present a simple strategy using multiscale superpixels to jointly detect salient object via low-rank optimisation. Specifically, we first build a multiscale superpixel pyramid and derive the corresponding saliency map by multimodal saliency features and priors at each single scale. Then, we use joint low-rank analysis of multiscale saliency maps to obtain a more reliable and adaptively-fused saliency map, which properly takes all scales saliency into account. We further propose a GMM generative co-saliency prior to enable the above approach to detect co-salient objects from multiple images. Extensive experiments on benchmark datasets validate the effectiveness and superiority of the proposed saliency and co-saliency detector over state-of-the-arts.

Saliency model-based face segmentation and tracking in head-and-shoulder video sequences

Video saliency detection aims to pop out the most salient regions in every frame of a video. Up to now, many efforts have been made from various aspects for video saliency detection. Unfortunately, the existing video saliency models are very likely to fail in challenging videos with complicated motions and complex scenes. Therefore, in this paper, we propose a novel framework to improve the saliency detection results generated by existing video saliency models. The proposed framework consists of three key steps including localized estimation, spatiotemporal refinement, and saliency update. Specifically, the initial saliency map of each frame in a video is first generated by using an existing saliency model. Then, by considering the temporal consistency and strong correlation among adjacent frames, the localized estimation models, which are generated by training the random forest regressor within a local temporal window, are employed to generate the temporary saliency map. Finally, by taking the appearance and motion information of salient objects into consideration, the spatiotemporal refinement step is deployed to further improve the temporary saliency map and generate the final saliency map. Furthermore, such an improved saliency map is then utilized to update the initial saliency map and provide reliable cues for saliency detection in the next frame. The experimental results on four challenging datasets demonstrate that the proposed framework is able to consistently and significantly improve the saliency detection performance of various video saliency models, thereby achieving the state-of-the-art performance.

The effectiveness of depth information in saliency detection has been fully proved. However, it is still worth exploring how to utilize the depth information more efficiently. Erroneous depth information may cause detection failure, while non-salient objects may be closer to the camera which also leads to erroneously emphasis on non-salient regions. Moreover, most of the existing RGB-D saliency detection models have poor robustness when the salient object touches the image boundaries. To mitigate these problems, we propose a multi-stage saliency detection model with the bilateral absorbing Markov chain guided by depth information. The proposed model progressively extracts the saliency cues with three level (low-, mid-, and high-level) stages. First, we generate low-level saliency cues by explicitly combining color and depth information. Then, we design a bilateral absorbing Markov chain to calculate mid-level saliency maps. In mid-level, to suppress boundary touch problem, we present the background seed screening mechanism (BSSM) for improving the construction of the two-layer sparse graph and better selecting background-based absorbing nodes. Furthermore, the cross-modal multi-graph learning model (CMLM) is designed to fully explore the intrinsic complementary relationship between color and depth information. Finally, to obtain a more highlighted and homogeneous saliency map in high-level, we structure a depth-guided optimization module which combines cellular automata and suppression-enhancement function pair. This optimization module refines the saliency map in color space and depth space, respectively. Comprehensive experiments on three challenging benchmark datasets demonstrate the effectiveness of our proposed method both qualitatively and quantitatively.

Intelligent surveillance systems enable secured visibility features in the smart city era. One of the major models for pre-processing in intelligent surveillance systems is known as saliency detection, which provides facilities for multiple tasks such as object detection, object segmentation, video coding, image re-targeting, image-quality assessment, and image compression. Traditional models focus on improving detection accuracy at the cost of high complexity. However, these models are computationally expensive for real-world systems. To cope with this issue, we propose a fast-motion saliency method for surveillance systems under various background conditions. Our method is derived from streaming dynamic mode decomposition (s-DMD), which is a powerful tool in data science. First, DMD computes a set of modes in a streaming manner to derive spatial–temporal features, and a raw saliency map is generated from the sparse reconstruction process. Second, the final saliency map is refined using a difference-of-Gaussians filter in the frequency domain. The effectiveness of the proposed method is validated on a standard benchmark dataset. The experimental results show that the proposed method achieves competitive accuracy with lower complexity than state-of-the-art methods, which satisfies requirements in real-time applications.

In this paper, we propose a novel method for automatic object segmentation from natural scene images. This method is based on saliency map, mean shift segmentation and level set method. First, a histogram based contrast method is used to generate the saliency map of the input image. Second, the input image is segmented into clusters using mean shift. Based on the saliency map, segmented clusters are classified into background and foreground clusters. After that, an initial contour for level set method is determined by applying morphological erosion on foreground clusters. Finally, the level set method is used to evolve the initial contour and find the object of interest. The experimental results have shown that our proposed method is not only able to replace manual labeling of initial contour in level set method but also able to yield more accurate segmentation results than previous segmentation approaches.

Salient object detection is a process of extracting an object which is visually attractive from a single image or a video. As a powerful technique for automatic image or video segmentation, saliency detection has been focused and studied recently. In this paper, we propose a novel method for salient object detection without training or learning-based techniques. The proposed framework consists of two major steps, the generation of saliency map candidates and the selection of an optimal saliency map. To generate saliency map candidates, prior maps based on combinations of RGB color components are proposed. To select the optimal saliency map among the candidates, we propose a compactness measure, which evaluates the degree to which the generated saliency maps show objects. As a result, among recent works on saliency detection, our saliency detection method achieves the highest performance in terms of saliency detection.

Saliency detection is a technique for automatically extracting regions of interest from the background and has been widely used in the computer vision field. This study proposes a simple and effective saliency detection method combining color contrast and hash fingerprint. In our solution, the input image is segmented into nonoverlapping superpixels, so as to perform the saliency detection at the region level to reduce computational complexity. A background optimization selection is used to construct an accurate background template. Based on this, a saliency map that highlights the whole salient region is obtained by estimating color contrast. Besides, another saliency map that enhances the salient region while restraining the background is also generated through hash fingerprint matching. Ultimately, the final saliency map can be obtained by fusing the two saliency maps. Comparing the performance with other methods, the proposed algorithm works better even in the presence of complex background or very large salient regions.

Eye tracking is a technique whereby an individual s eye movements are measured so that the researcher knows both where a person is looking at any given time and the sequence in which their eyes are shifting from one location to another. In the research Œeldof human-computer interaction (HCI), human intentions can be recognized by analyzing gaze information. On the other hand, people s intentions are largely inceuenced by a number of objective factors. Therefore, we introduce saliency map as a basis for intention recognition. In the process of saliency map calculation, fuzzy neural network is employed in order to reœect individual characteristics importance of images. In this study, we propose a method realized the intention recognition by combining gaze information and saliency map of images. To verify the effectiveness of proposed method, an interactive teaching system is constructed based on eye tracking and saliency map by using an omnidirectional wheelchair.