This article designs an underwater binocular measurement system combining binocular vision and multicolor structured light, for the problem of autonomous grasping by underwater robots. In our solution, multiple colored stripes of structured light are projected on the surface of the object to be measured at once without the scanning process and, thus, have the advantages of high measurement accuracy, efficiency, stability, and reliability, which could realize the survey and positioning of underwater targets and guide the robotic arm to grasp the underwater targets autonomously. In this article, an underwater binocular measurement model with nonparallel and non-co-refractive surfaces is established by tracing the propagation path of light in water, and a multicolor structured light array is used to provide active visual features for the underwater object to be measured by projecting the multicolor structured light array, avoiding the limitation of the center point of monochromatic structured light, and the object could be at any position in the binocular field of view. Then, the laser strip images were separated from the background and segmented by the HSV double-threshold segmentation method; the color light stripes segmented from the left and right images were matched corresponding to their color information and position information. Finally, the feature points required for measurement are extracted from the laser stripe images taken by the left and right cameras to achieve a fast underwater survey, and through analyzing the experimental data and observing the object 3-D reconstruction effect, the effectiveness and accuracy of the underwater binocular measurement model and the underwater binocular matching algorithm established in this article are proved.
8 May 2026 in Journal of the Brazilian Society of Mechanical Sciences and Engineering
In recent years, the field of virtual reality has developed by leaps and bounds. As an important intelligent technology in the process of virtual life, voice driven virtual face animation can use a piece of voice to create real and natural face animation of virtual characters. It is widely used in the fields of intelligence education, virtual anchor, digital entertainment and so on, which meets the needs of human-computer interaction in the field of virtual reality. In order to further promote the technological breakthrough and intelligence in the process of virtual face animation generation, this paper compares and analyzes the key problems of voice driven virtual face animation, combs and summarizes the core technologies such as voice visual mapping, face model construction and virtual face animation generation, and looks forward to the future research and development direction of voice driven face animation, It is of great significance to the further research and development of speech driven facial animation.
27 March 2026 in Measurement Science and Technology
In this paper, an adaptive method for tire tread depth measurement based on binocular line laser is proposed and implemented. Important challenges such as the interference of wear indicators, the interference of lateral tread grooves, and the adaptability to tires with different specifications are addressed. First, a binocular 3D laser reconstruction platform was set up to capture complete point cloud data of the tire crown. To overcome the scattering and non-uniform brightness of the laser stripes on the tire crown, a sub-pixel algorithm for extracting the centerline of the stripe was developed based on the normal convergence of offset points, which greatly improved the stability and accuracy compared with the gray centroid method and Steger algorithm (by about 25.3% and 12.2%, respectively). In processing the point cloud, a multi-level segmentation strategy for wear indicators was developed, combining geometric feature constraints and region growing, and an adaptive DBSCAN with distance thresholds was used to remove lateral groove points and localize points on the tread. To adapt to different tire specifications, an adaptive algorithm for extracting grooves on the tire base was proposed based on the cross-sectional width of the tire. The experiments used four passenger-car tires of different specifications as validation objects, and these four specifications are all widely used mainstream tire specifications. The absolute error between the measurement results and those obtained using a manual depth gauge is less than 0.2 mm, and the measurement efficiency is improved by 2.5 times compared with the manual method. The method proposed in this paper meets the requirements of industrial online inspection in terms of accuracy, robustness, and generality.