This study presents an approach for estimating the fisheye camera parameters using three vanishing points corresponding to three sets of mutually orthogonal parallel lines in one single image. The authors first derive three constraint equations on the elements of the rotation matrix in proportion to the coordinates of the vanishing points. From these constraints, the rotation matrix is calculated under the assumption of the image centre known. The experimental results with synthetic images and real fisheye images validate this method. In contrast to the existing methods, the authors method needs less image information and does not know the three‐dimensional reference point coordinates.
At present, the sensitivity phase of low-frequency accelerometer is commonly calibrated by time synchronization (TS), which needs to strictly align its input excitation acceleration signal and output signal in the time domain. However, TS is very difficult to be implemented and has severely restricted the improvement of the measurement accuracy. A novel calibration method that combines the monocular vision method and time-spatial synchronization technique is investigated to achieve the high-accuracy sensitivity phase calibration. The sensitivity phase is accurately calibrated by determining the aligned spatial position between the excitation acceleration signal and the output signal with the monocular vision method. The sensitivity magnitude can also be simultaneously calibrated. Experimental results show that the calibrated sensitivity phase and magnitude by the investigated method agree well with those by the laser interferometry in the range from 0.3 Hz to 2 Hz. The calibration accuracy of the investigated method is especially superior to that of the laser interferometry in the range from 0.01 Hz to 0.3 Hz.
The calibration of acoustic emission sensor is important for acoustic emission quantitative evaluation. To carry out the sensor calibration, one way is to find a reference acoustic emission source, such as a falling solid ball, a fracturing pencil lead, with energy evaluation to obtain the characteristics of the reference source. One way is to use a reference sensor to measure the pressure or the normal velocity at the surface, including the capacitive transducer and the laser interferometer. The other is using reciprocity method. Frequency characteristics of amplitude of absolute sensitivity of both the Rayleigh and longitudinal waves could be determined by purely electrical measurement without the use of mechanical sound sources of reference transducers. In this paper, the progress of the methods of the AE sensor calibration was given. The principles and the merits and faults of each method are discussed. In this paper the rapid calibration method by pulse exciting transducer was also discussed in different wave mode, as well as Face to Face method.
Abstract Small object detection has always been one of the most challenging tasks in the computer vision. Up to now, a prior bounding box is often applied to Unmanned Aerial Vehicle (UAV) image object detection. However, anchors need to be pre-set and not optimal for training data in many object detection algorithms. In 2022, the Diffusion Model was introduced in object detection method, in which the random boxes are employed. Inspired by this approach and the characteristics of UAV images, we find the great potential of diffusion models in UAV image detection and propose a more reasonable Decoupled Region of Interest Pooling Feature Diffusion Network. First of all, a more rational decoupled region of interest pooling(DRIP) feature extraction module has been designed, which decouples the feature extraction process between different scales, to make full use of the features at each level of the pyramid. Our approach eliminates the negative effects of unreasonable bounding box assignments, thereby enhancing the overall performance. Secondly, we propose a high-resolution scale-varying robust backbone(HSRB), where we architect the convolution module in the backbone using atrous convolution with switchable atrous rates and Pixel-Shuffle upsampling to mitigate the negative effects of scale variation and downsampling. Finally,loss functions with normalized Wasserstein distance (NWD) terms are applied, NWD is led into measuring the similarity between the prediction box and the ground truth box. The purpose is to eliminate the influence of positional sensitivity on the matching between the predicted box and the ground truth box.The optimal results of 27.91% mAP on the VisDrone dataset and 8.42% mAP on the TinyPerson dataset demonstrate the effectiveness of the proposed model.
In recent years, low frequency vibration measurement is being widely concerned in many applications, because low frequency vibration usually introduces a strong influence. The low frequency vibration is commonly measured by the laser interferometry, requiring a complicated system and a high cost laser interferometer, and its flexibility in field vibration measurement is poor; or the method using vibration transducer, requiring a transducer with known sensitivity, and its measurement precision is not high. In this paper, we propose a novel method based on single camera, which achieves low frequency vibration measurement via collecting and processing sufficient frame sequence images. The proposed method is compared with the heterodyne interferometry by measuring the vibration displacements of a long-stroke shaker simultaneously. Experimental results show the proposed method realizes <1% vibration displacement measurement precision at frequencies between 0.05 Hz-5 Hz.
This paper utilizes firstly both a scanning device and an optic fiber hydrophone to establish a measurement system, and then proposes the parameter measurement of the focused transducer based on edge detection of the visualized acoustic data and curve fitting. The measurement system consists of a water tank with wedge absorber, stepper motors driver, system controller, a focused transducer, an optic fiber hydrophone and data processing software. On the basis of the visualized processing for the original scanned data, the −3 dB beam width of the focused transducer is calculated using the edge detection of the acoustic visualized image and circle fitting method by minimizing algebraic distance. Experiments on the visualized ultrasound data are implemented to verify the feasibility of the proposed method. The data obtained from the scanning device are utilized to reconstruct acoustic fields, and it is found that the −3 dB beam width of the focused transducer can be predicted accurately.
Most of the existing object detection systems adopt 3x3 convolution kernels for feature extraction, which leads to a problem that the receptive field of the features extraction net is always 3 X • Network features are not rich enough and lack accurate learning of features with pixel size not 3 x • To solve this problem, convolution with different kernel sizes is introduced for feature extraction. However, a large convolution kernel may lead to a rapid increase in the Parameters and FLOPs. In this paper, we propose an object detection network based on depth-wise convolution and multi-scale feature fusion (YOLODM-Net). Specifically, a feature extraction module named multi-scale feature fusion (MSFF) block is constructed, which uses depth-wise convolution of different kernel sizes to extract features and mixes them to enrich learning contents. In addition, we propose a multi-scale spatial attention module based on the Efficient Channel Attention (ECA) module. In this module, multi-scale information is added to make the extracted features more fine-grained. The proposed method was evaluated on the VOC2007 dataset and compared with the previous methods. The mAP of the model is better than that of the YOLOv7, YOLOx, etc. And the Parameters and FLOPs are also improved.
Low brightness contrast and grey level discontinuities of the ultrasonic liver image make it difficult to segment the object and the background and to extract the edges of the object using the global optimal threshold method. In this paper, we investigate a local optimal threshold method for the segmentation of ultrasound liver image. First of all, the distributed energy of the ultrasound liver image is estimated in the proposed liver segmentation. Then, the polynomials are fitted from the distributed energy data and a peak zone is defined from the minimum of the fitted polynomials. Finally, a few blocked images are divided from the number of the peak zones. Furthermore, multiple local optimal thresholds are obtained from the blocked images using Otsu's method, and the ultrasonic liver image is segmented according to all local optimal thresholds. Experimental results validate the segmentation and edge detection of liver in the ultrasound images.
'/%3e%3cuse xlink:href='%23a' transform='rotate(23.036 .093 25.536)'/%3e%3cuse xlink:href='%23a' transform='rotate(45.87 1.273 16.18)'/%3e%3cuse xlink:href='%23a' transform='rotate(69.945 .996 12.078)'/%3e%3cuse xlink:href='%23a' transform='rotate(20.66 -19.689 31.932)'/%3e%3c/svg%3e)