Abstract: A machining parameter automatic generation system includes a geometric data capturing module, a feature recognition learning network and a machining parameter learning network. The geometric data capturing module captures a geometric shape of a workpiece to generate a candidate feature list. The feature recognition learning network trains the candidate feature list according to a first neural network model to obtain an applicable feature list. The machining parameter learning network trains the applicable feature list and the candidate machining parameter according to a second neural network model to obtain an applicable machining parameter. The applicable machining parameter is used to generate a machining program, and the machining program is read by a machine tool for processing.

Abstract: A machining parameter automatic generation system includes a geometric data capturing module, a feature recognition learning network and a machining parameter learning network. The geometric data capturing module captures a geometric shape of a workpiece to generate a candidate feature list. The feature recognition learning network trains the candidate feature list according to a first neural network model to obtain an applicable feature list. The machining parameter learning network trains the applicable feature list and the candidate machining parameter according to a second neural network model to obtain an applicable machining parameter. The applicable machining parameter is used to generate a machining program, and the machining program is read by a machine tool for processing.

Abstract: A calibration method applicable for an automation machining apparatus includes building a first stereoscopic characteristic model corresponding to an object, obtaining a stereoscopic image of the object, building a second stereoscopic characteristic model corresponding to the object based on the stereoscopic image, obtaining at least one error parameter corresponding to the second stereoscopic characteristic model by comparing the second stereoscopic characteristic model with the first stereoscopic characteristic model, and calibrating a machining parameter of the automation machining apparatus based on the at least one error parameter.

Abstract: A method for vision machine inspection comprises providing depth information of a target acquired by an image capturing system, determining real-time three-dimensional information of a target object in a predetermined inspecting area based on depth information of at least one real-time image of the target. The method further comprises projecting color pixel information of a real-time color image of the target object to a three-dimensional virtual model based on the real-time three-dimensional information. The real-time color image may be acquired by a color camera system. The method further comprises generating a color three-dimensional virtual model. The color three-dimensional virtual model may comprise the color pixel information.

Abstract: A calibration method applicable for an automation apparatus includes building a first stereoscopic characteristic model corresponding to an object, obtaining a stereoscopic image of the object, building a second stereoscopic characteristic model corresponding to the object based on the stereoscopic image, obtaining at least one error parameter corresponding to the second stereoscopic characteristic model by comparing the second stereoscopic characteristic model with the first stereoscopic characteristic model, and calibrating a processing parameter of the automation apparatus based on the at least one error parameter.

Abstract: A method and an apparatus for reconstructing a three dimensional model of an object are provided. The method includes the following steps. A plurality of first depth images of an object are obtained. According to a linking information of the object, the first depth images are divided into a plurality of depth image groups. The linking information records location information corresponding to a plurality of substructures of the object. Each depth image group includes a plurality of second depth images, and the substructures correspond to the second depth images. According to the second depth image and the location information corresponding to each substructure, a local module of each substructure is built. According to the linking information, the local models corresponding to the substructures are merged, and the three-dimensional model of the object is built.

Abstract: A method for vision machine inspection comprises providing depth information of a target acquired by an image capturing system, determining real-time three-dimensional information of a target object in a predetermined inspecting area based on depth information of at least one real-time image of the target. The method further comprises projecting color pixel information of a real-time color image of the target object to a three-dimensional virtual model based on the real-time three-dimensional information. The real-time color image may be acquired by a color camera system. The method further comprises generating a color three-dimensional virtual model. The color three-dimensional virtual model may comprise the color pixel information.

Abstract: A system and method for determining individualized depth information in an augmented reality scene are described. The method includes receiving a plurality of images of a physical area from a plurality of cameras, extracting a plurality of depth maps from the plurality of images, generating an integrated depth map from the plurality of depth maps, and determining individualized depth information corresponding to a point of view of the user based on the integrated depth map and a plurality of position parameters.

Abstract: A method and an apparatus for reconstructing a three dimensional model of an object are provided. The method includes the following steps. A plurality of first depth images of an object are obtained. According to a linking information of the object, the first depth images are divided into a plurality of depth image groups. The linking information records location information corresponding to a plurality of substructures of the object. Each depth image group includes a plurality of second depth images, and the substructures correspond to the second depth images. According to the second depth image and the location information corresponding to each substructure, a local module of each substructure is built. According to the linking information, the local models corresponding to the substructures are merged, and the three-dimensional model of the object is built.

Abstract: A depth detection method includes the following steps. First, first and second video data are shot. Next, the first and second video data are compared to obtain initial similarity data including r×c×d initial similarity elements, wherein r, c and d are natural numbers greater than 1. Then, an accumulation operation is performed, with each similarity element serving as a center, according to a reference mask to obtain an iteration parameter. Next, n times of iteration update operations are performed on the initial similarity data according to the iteration parameter to generate updated similarity data. Then, it is judged whether the updated similarity data satisfy a character verification condition. If yes, the updated similarity data is converted into depth distribution data.

Abstract: A surface treatment method is provided. A surface treatment method comprising steps of providing a first and a second conductors; applying an electrical field between the first and the second conductors; enclosing the first and the second conductors with a material whose viscosity is varied with an intensity of the electrical field; actuating the first and the second conductors such that the first and the second conductors are in a relative motion with respect to each other; and varying the intensity of the electrical field.

Abstract: A method for recognizing three-dimensional control points and a computer readable medium using the same are disclosed. The method for recognizing three-dimensional control points comprises the following steps. A depth information item corresponding to an image captured by an image capturing apparatus is received. A three-dimensional block information item corresponding to a three-dimensional block is generated according to the depth information. At least one reference plane is generated according to the depth information. At least one connection group is generated according to the three-dimensional block information and the reference plane. A three-dimensional block nearest to the image capturing apparatus is selected as a control point from the connection group.

Abstract: A system for estimating a location of an occluded skeleton, a method for estimating a location of an occluded skeleton and a method for reconstructing an occluded skeleton are provided. The method for estimating a location of an occluded skeleton comprises the following steps: Firstly, a trace of a reference central point of a body is estimated according to a plurality of continuously moving images. Next, a human movement state is estimated according to the trace and a motion information of the continuously moving images free of skeleton occlusion. Then, a possible range of the occluded skeleton for maintaining human balance is calculated according to the human movement state. Afterwards, a current motion level of the occluded skeleton is predicted according to a historic motion information of the occluded skeleton. Lastly, the location of the occluded skeleton is estimated according to the current motion level and the possible range.

Abstract: A system for estimating a location of an occluded skeleton, a method for estimating a location of an occluded skeleton and a method for reconstructing an occluded skeleton are provided. The method for estimating a location of an occluded skeleton comprises the following steps: Firstly, a trace of a reference central point of a body is estimated according to a plurality of continuously moving images. Next, a human movement state is estimated according to the trace and a motion information of the continuously moving images free of skeleton occlusion. Then, a possible range of the occluded skeleton for maintaining human balance is calculated according to the human movement state. Afterwards, a current motion level of the occluded skeleton is predicted according to a historic motion information of the occluded skeleton. Lastly, the location of the occluded skeleton is estimated according to the current motion level and the possible range.

Abstract: A method for recognizing three-dimensional control points and a computer readable medium using the same are disclosed. The method for recognizing three-dimensional control points comprises the following steps. A depth information item corresponding to an image captured by an image capturing apparatus is received. A three-dimensional block information item corresponding to a three-dimensional block is generated according to the depth information. At least one reference plane is generated according to the depth information. At least one connection group is generated according to the three-dimensional block information and the reference plane. A three-dimensional block nearest to the image capturing apparatus is selected as a control point from the connection group.

Abstract: A depth detection method includes the following steps. First, first and second video data are shot. Next, the first and second video data are compared to obtain initial similarity data including r×c×d initial similarity elements, wherein r, c and d are natural numbers greater than 1. Then, an accumulation operation is performed, with each similarity element serving as a center, according to a reference mask to obtain an iteration parameter. Next, n times of iteration update operations are performed on the initial similarity data according to the iteration parameter to generate updated similarity data. Then, it is judged whether the updated similarity data satisfy a character verification condition. If yes, the updated similarity data is converted into depth distribution data.

Abstract: A machining fluid is provided. A machining fluid for one of an electrical discharge machining process and an electrical discharge machining-polishing process, comprising a polymolecular powder, a hard particle and a carrier liquid, wherein a concentration of the polymolecular powder is lower than 500 g/L.