70results about How to "Expand the border" patented technology

The invention relates to a viewing system and a method for detecting an object of interest in a sequence of images (IS). Said object of interest is detected by first locating localizers related to said object of interest and by locating borders (BL) related to said object of interest using the location (LI, LZ) of said localizers. The viewing system according to the invention is able to produce a sequence of enhanced images in which the object of interest is enhanced, to measure some characteristics and to build a three dimensional representation of said object of interest. The viewing system is also able to register and combine said sequence of enhanced images with a sequence of reference images.

Disclosed is a jacquard raising needle machine. The machine comprises a machine frame, and the machine frame comprises a base, left and right box frames and a top beam; a double shaft needle beam driving mechanism is arranged on the top beam, the needle beam is connected with the double shaft needle beam driving mechanism, a needle plate is arranged on the side, facing downwards, of the needle beam, and raising needles are arranged on the needle plate; a raising mechanism is arranged on the part between the left and right box frames; a brush curtain power transmission mechanism is arranged onthe base and connected with the raising mechanism; a net stripping plate is connected with a net stripping plate adjustment mechanism, and the net stripping plate lifting adjustment mechanism is arranged below the top beam. The machine is characterized in that the machine further comprises brush curtain frame left and right lifting adjustment mechanisms and a brush curtain frame traction mechanism, the brush curtain frame left and right lifting adjustment mechanisms are arranged on the left and right box frames, the brush curtain frame traction mechanism is arranged on the left end of the baseand connected with the brush curtain left lifting adjustment mechanism, and the raising mechanism is connected with the part between the brush curtain frame left and right lifting adjustment mechanisms. The machine has the advantages that the level of non-woven products is improved, and the edge effect of suede patterns and the overall effect of the suede patterns are improved.

The invention provides a digital audio-visual system, a management system and a method for user interaction in the digital audio-visual system. Specifically, a system server and user terminals are provided, the user terminals are connected to the system server through a network, and the user terminals are terminal equipment logged in by users; the first user establishes a first user group on the system server, and the first user group is associated with a first digital audio-visual location; other users join in the first user group, audio-visual recreational activity information in the first digital audio-visual location acts as a group message and is sent to the first user group, and the users receive or query the group message through the logged-in user terminals. According to the invention, a technology platform for communicating the internet information world and realistic digital audio-visual recreational activities is established, the boundary of digital audio-visual locations for carrying out the digital audio-visual recreational activities are greatly expanded, and the degree of interaction among the users is improved.

The invention provides a method and device for processing data category imbalance, electronic equipment and a storage medium. The invention relates to the field of big data processing of cloud technology. The method comprises the steps of determining M nearest neighbor samples based on mutual information between minority class samples Xi and each neighbor sample of the Xi; determining a mutual information weight of the Xij (near) based on the mutual information between the Xi and the Xij (near); determining the weight Wij (near) of the Xij (near) based on the type of the Xij (near) and the mutual information weight of the Xij (near); determining the number Nj of minority class samples to be inserted between the Xi and the Xij (near) based on the Wij (near) and the class imbalance multiplying power N; and inserting Nj new samples between the Xi and the Xij (near). The problem of data category imbalance is solved in a mode of fusing mutual information and SMOTE, and the classification performance of the SMOTE can be improved.

The embodiment of the invention provides a joint low-rank constraint cross-view discrimination subspace learning method and device. The method comprises the following steps: defining an objective function of a double-low rank discrimination subspace learning formula; adopting a supervised regularization term as a strong constraint condition, and re-drawing up the target function; adding a joint heterogeneous regularization term to re-draw up the target function; dividing the image data set into a test set and a training set; utilizing the training set to solve the value of each variable when the objective function value is minimized; solving the objective function to obtain a feature subspace; and projecting the test set through the feature subspace to obtain all features of all types of images in the data set, and finally obtaining the recognition rate of the data set through a classifier. According to the method, heterogeneous regularization items are combined as constraints to construct discriminant items for feature learning, isomorphic and heterogeneous information of a sample can be projected to a subspace to be used in a discriminant learning model of an image recognition and classification task, and model adaptability and robustness are promoted.