Aiming at the problem that stereo matching accuracy is easily affected by noise and amplitude distortion, a stereo matching algorithm based on HSV color space and improved census transform is proposed. In the cost calculation stage, the color image is first converted from RGB space to HSV space; moreover, the hue channel is used as the matching primitive to establish the hue absolute difference (HAD) cost calculation function, which reduces the amount of calculation and enhances the robustness of matching. Then, to solve the problem of the traditional census transform overrelying on the central pixel and to improve the noise resistance of the algorithm, an improved census method based on neighborhood weighting is also proposed. Finally, the HAD cost and the improved census cost are nonlinearly fused as the initial cost. In the aggregation stage, an outlier elimination method based on confidence interval is proposed. By calculating the confidence interval of the aggregation window, this paper eliminates the cost value that is not in the confidence interval and subsequently filters as well as aggregates the remaining costs to further reduce the noise interference and improve the matching accuracy. Experiments show that the proposed method can not only effectively suppress the influence of noise, but also achieve a more robust matching effect in scenes with changing exposure and lighting conditions.
1 April 2005 in Japanese Journal of Applied Physics
We report a theoretical study on the electronic structures of indium oxide (In2O3) and indium tin oxide (ITO) carried out using our original tight-binding quantum chemical calculation program “Colors”, which is over 5,000 times faster than the conventional first-principles quantum chemical calculation method. The calculated band gap of In2O3 is in good agreement with the experimental results, although the value obtained by conventional first-principles calculation is less than half the experimental one. The electronic structures of In2O3 calculated by our tight-binding method are consistent with those obtained by first-principles calculations. Furthermore, the doping of tin atoms into In2O3 increased the band gap, which is also in good agreement with the experimental tendency. Hence, we confirmed that our tight-binding quantum chemical calculation method was very effective for investigation and predicting the electronic structures of In2O3 and ITO crystals with high accuracy and reliability in spite of its high calculation speed.
At present, visual saliency prediction algorithms have been developed more and more mature, but most of the current saliency prediction algorithms are aimed at natural images. Due to the inconsistency of elements and features between natural images and advertising images, the existing saliency prediction algorithms show poor robustness and low inference speed to advertising images, which severely limits its commercial application in advertising design and evaluation. In view of this, a saliency prediction algorithm for advertisement images is proposed in this paper. In the feature extraction stage, two text candidate regions based on intensity feature and improved MESR algorithm are first obtained and further integrated to produce a two-dimensional text confidence score. Meanwhile, a saliency confidence score is also obtained by an improved natural image saliency prediction network. Then, the score level fusion strategy was adopted to fuse the two confidence scores to get the final saliency prediction map. The experimental results show that the proposed model has good accuracy and robustness in advertising images, as well as the most remarkable inference speed, which can meet the demand for real-time performance of advertising image saliency prediction, leading to great practical and commercial value.
2 February 2026 in IEEE Transactions on Multimedia
Self-supervised depth estimation has been widely ap plied in indoor environments. However, the presence of numerous objects and complex structural boundaries often leads existing methods to generate blurred or imprecise depth edges. To address this challenge, we propose FGDepth, a framework designed to enhance depth estimation through fine-grained boundary perception. Firstly, we introduce an SR (Super Resolution) auxiliary training branch that shares feature layers with the encoder of the depth estimation network. By utilizing the powerful detail recovery capabilities of the SR task, we improve the depth network's sensitivity to indoor object boundaries. Notably, the SR branch is used only during training, ensuring no added computational cost during inference. As far as we know, we are the first to employ the SR task as an auxiliary method for indoor self-supervised depth estimation. Second, we observe that outdoor scenes display significant depth variations due to their broader depth range, whereas indoor scenes typically lack clear boundary distinctions in background areas. This is because distant objects in indoor settings often appear as continuous surfaces with similar depths. This characteristic necessitates careful mitigation of background depth uniformity interference when enhancing depth boundaries in indoor scenes. Therefore, we design a depth-adaptive object mask to provide target object boundary information and use a triplet loss to align these differences with the depth map. Experimental results on three benchmark datasets show that our method outperforms existing approaches. We also conduct ablation studies to validate the contributions of each component.
23 April 2025 in IEEE Transactions on Instrumentation and Measurement
Indoor depth measurement is widely used in technologies such as virtual reality and augmented reality. However, indoor scenes are typically captured with handheld cameras, resulting in more complex and unpredictable variations between frames. Self-supervised depth estimation relies on frame-to-frame projection for self-constraint, and inaccurate pose predictions between frames significantly hinder depth estimation in indoor environments. To address this issue, PRDepth is proposed, a self-supervised pose refinement method tailored for indoor environments. PRDepth introduces a Pose Reconstruction Iterative Module that refines multi-frame pose decomposition and reconstruction. By leveraging contextual information from intermediate frames, it mitigates estimation errors caused by large rotations and reduces training errors due to inaccurate rotations, leading to more precise pose predictions between frames. Additionally, to enhance the information exchange and context integration capabilities of the depth network, PRDepth features a depth-weighted incentive module, which includes a Global Depth Enhancement Module in the encoder-decoder and a Weight-Adaptive Incentive Module in the decoder. The Global Depth Enhancement Module improves the network’s ability to extract depth information in complex scenes by interacting with global cross-dimensional data. An attention-guided mechanism is adopted by the Weight-Adaptive Incentive Module to aggregate multi-scale feature information and assign adaptive weights to different features, ensuring efficient global context fusion and suppression of redundant information. Experimental results demonstrate that our method significantly outperforms existing state-of-the-art self-supervised monocular depth estimation techniques for indoor scenes. Extensive ablation studies are conducted on each module of PRDepth. PRDepth demonstrates precise depth estimation and robust generalization across indoor datasets, including NYUv2, 7-Scenes, ScanNet, and InteriorNet.
1 October 2021 in Mathematical Problems in Engineering
Target tracking is currently a hot research topic in machine vision. The traditional target tracking algorithm based on the generative model selects target features manually, which has a simple structure and fast running speed, but it cannot meet the requirements of algorithm accuracy in complex scenes. Compared with traditional algorithms, due to the good performance, the tracking method based on full convolutional network has become one of the important methods of target tracking. However, the RPN-based Siamese network lacks positional reliability when predicting the target area. Aiming at the low tracking accuracy of the RPN-based Siamese network, this paper proposes an improved framework model named IoU-guided SiamRPN (IG-SiamRPN). In the proposed IG-SiamRPN, the IoU-guided branch is first constructed and sample pairs are generated through data augmentation. Then, the Jittered RoI is constructed to train the network to realize the direct prediction of the localization confidence of the candidate area. Subsequently, a target selection method based on predicted IoU scores is proposed, which uses predicted IoU scores instead of classification scores to optimize the target decision strategy of the Siamese network. Finally, an optimization-based fine-tuning method for the Siamese network frame is proposed, which solves the problem of location degradation and improves the performance of the algorithm. Compared with other state-of-the-art target tracking algorithms, experimental results on popular databases demonstrate that the proposed IG-SiamRPN can achieve better performance in both tracking accuracy and robustness.