Profile measurement is a key technical enabler in the manufacturing of highly curved freeform surfaces due to their complex geometrical shape. A current optical probe was used to measure nearly rotary freeform surfaces with the help of one rotation axis, because the probe needs to measure along the normal vector of the surface under the limitation of the numerical aperture (NA). This kind of measuring system generally has a high cost due to the high-precision, multi-axis platform. In this paper, we propose a low-cost, dual-axis rotation scanning method for a highly curved freeform surface with an arbitrary shape. The optical probe can scan the surface profile while always keeping consistent with the normal vector of the measuring points with the help of the double rotation axis. This method can adapt to the changes in curvature in any direction for a highly curved freeform surface. In addition, the proposed method provides a system error calibration technique for the rotation axis errors. This technique can be used to avoid the dependence of the measuring system on the high-precision platform. The three key system errors that affect the measurement accuracy such as installation error of the B-axis, A-axis, and XZ perpendicularity error are first analyzed through establishing an error model. Then, the real error values are obtained by the optimal calculation in the calibration process. Finally, the feasibility of the measurement method is verified by measuring one cone mirror and an F-theta mirror and comparing the results to those obtained using commercial equipment. The maximum measurable angle of the system is ±90°, the maximum measurable diameter is 100 mm, and the measurement accuracy of the system reaches the micron level in this paper.
Domain adaptation methods face performance degradation in object detection, as the complexity of tasks require more about the transferability of the model. We propose a new perspective on how CNN models gain the transferability, viewing the weights of a model as a series of motion patterns. The directions of weights, and the gradients, can be divided into domain-specific and domain-invariant parts, and the goal of domain adaptation is to concentrate on the domain-invariant direction while eliminating the disturbance from domain-specific one. Current UDA object detection methods view the two directions as a whole while optimizing, which will cause domain-invariant direction mismatch even if the output features are perfectly aligned. In this paper, we propose the domain-specific suppression, an exemplary and generalizable constraint to the original convolution gradients in backpropagation to detach the two parts of directions and suppress the domain-specific one. We further validate our theoretical analysis and methods on several domain adaptive object detection tasks, including weather, camera configuration, and synthetic to real-world adaptation. Our experiment results show significant advance over the state-of-the-art methods in the UDA object detection field, performing a promotion of $10.2\sim12.2\%$ mAP on all these domain adaptation scenarios.
Generally different websites have different web page structures, which would heavily affect the extraction quality when the web content is automatically collected. On the basis of a statistical analysis on content features and structure characteristics of News domain web pages, this paper proposes a maximum continuous sum of text density (MCSTD) method to efficiently and effectively extract web content from different web pages. Firstly, web pages are preprocessed, and then the text density of texts are calculated. Finally, the web content is extracted using the proposed MCSTD method. Experimental results show that the extraction precision is over 95%, and the proposed approach is more efficient and easier to be implemented compared to traditional models. Additionally, our method has also been applied to the scenario of comparable corpora construction using extracted web resource.
Mathematical model and control strategy are essential for active thermal control of nano-satellites. This paper presents a 5-nodal thermal network for transient performance calculation of heat-pipe thermal system which governs the temperature environment inside the nano-satellites. Two predictive control algorithms including implicit generalized predictive self-tuning control (IGPSC) and dynamic matrix control (DMC) were presented and discussed for the close-loop control of nano-satellite's heat-pipe thermal system, their control effects were numerically investigated and compared under different thermal disturbances. The numerical investigation results suggest that both IGPSC and DMC methods are effective for the nano-satellite's active thermal control. However, the control effects under DMC are better than that under IGPSC for excellent temperature tracking ability, fast responses and small overshoots.
Objective The current study was conducted to determine whether peak bone mineral density (BMD) and obesity phenotypes are associated with certain LGR4 gene polymorphisms found in Chinese nuclear families with female children. Methods A total of 22 single nucleotide polymorphisms (SNPs) located in and around the LGR4 gene were identified in 1,300 subjects who were members of 390 Chinese nuclear families with female children. Then, BMD readings of the femoral neck, total hip, and lumbar spine as well as measurements of the total lean mass (TLM), total fat mass (TFM), and trunk fat mass were obtained via dual-energy X-ray absorptiometry. The quantitative transmission disequilibrium test was used to analyze the associations between specific SNPs and LGR4 haplotypes and peak BMD as well as between LGR4 haplotypes and TLM, percent lean mass, TFM, percent fat mass, trunk fat mass, and body mass index (BMI). Results Here, rs7936621 was significantly associated with the BMD values for the total hip and lumbar spine, while rs10835171 and rs6484295 were associated with the trunk fat mass and BMI, respectively. Regarding the haplotypes, we found significant associations between GAA in block 2 and trunk fat mass and BMI, between AGCGT in block 3 and total hip BMD, between TGCTCC in block 5 and femoral neck BMD, and between TACTTC in block 5 and both lumbar spine and femoral neck BMD (all P -values < 0.05). Conclusion Genetic variations of the LGR4 gene are related to peak BMD, BMI, and trunk fat mass.
'/%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)