45results about How to "Guaranteed naturalness" patented technology

The invention discloses a gait type identification method based on a three-axis acceleration sensor and a neural network. The method specifically comprises steps as follows: step 1), establishing a database of gait acceleration signals; step 2), performing the segmentation stage of the signals in the corresponding period; step 3), removing a gravity factor; step 4), performing the gait feature extraction stage; step 5), performing gait presorting stage; step 6), performing dimensionality reduction operation of a gait feature set; and step 7), performing specific gait identification stage. According to the method, gait features are screened with a staged MIV (mean impact value) method, the gait type identification work is performed in combination of a BP (back propagation) neural network, the extracted features are taken as input independent variables of the neural network, six gait types of sitting, standing, walking in a low speed, walking in a high speed, going upstairs and going downstairs are effectively identified sequentially through the gait presorting stage and the specific gait identification stage, and the method can have higher accuracy and reliability through enlarging of a gait data capacity range and the optimized design of the neural network.

The invention belongs to the field of computer vision, and provides good detail while improving image contrast, and the natural performance of the tone is kept while the noise is suppressed. An underwater image enhancement method based on structure-texture layering comprises the steps of firstly, carrying out color correction through histogram equalization, decomposing the image subjected to colorcorrection into a low-frequency structural layer and a high-frequency texture layer, leaving the noise on the texture layer, accurately estimating the transmission rate from the structural layer without noise based on a proposed fog line model to conduct enhancement processing, enhancing the texture layer by using a gradient residue minimization method, reconstructing the reinforced structural layer, the texture layer and the refined edge mask to form a final enhanced image through scaling, so that the problems that can not be solved through existing technology is solved. The method is mainlyapplied to underwater image enhancement and processing occasions.

The invention provides an interactive image processing method used for a video. The method comprises the following steps of extracting key frames from a video sequence; overlaying frames between adjacent key frames one by one to form a key frame cluster used for interactive marking; distinguishing the images of the key frames into a foreground area, a background area and an unknown area; carrying out spectral clustering and alpha value estimation on the key frames to obtain the sectional drawing result of the key frames; and finally, transferring the sectional drawing result of the key frames to the whole video sequence to obtain a final sectional drawing result. Because the foreground points are interactively marked on each key frame cluster, spatiotemporal information contained in the video sequence is fully utilized, and the naturalness and the intuition of user interaction are guaranteed on the basis of the user interface of the key frame, thereby conforming to the observation habit of a human visual system. Marking on the key frame clusters replaces independently marking on the key frames, and therefore the method has strong robustness for the situation with the big foreground object local action.

The invention discloses a gesture tracking recognition joint strategy method oriented to deskVR, At first, that hand gesture public data set ICVL is train offline by the convolution neural network, get a good robustness, models with high accuracy and fast recognition speed, and by testing the balance point Rate of accuracy and speed of the combined method with synthetic data, Then the depth imagesare captured by the depth camera in real time. After a series of preprocessing, the images are transferred into the gesture recognition method based on the fast model or the five-region convolution neural network model according to the proportion of the hand images, and the 3D information of the recognized joints is returned, and then the 3D human body model is mapped. The recognition method of the invention has good interactive function under the deskVR application environment, and performs well in recognition rate, operation speed, universality, accuracy and the like.

The invention relates to the technical field of electric car active sound generation, and discloses an electric car active sound generation system and a sound control method thereof. The sound controlmethod of the electric car active sound generation system comprises the following steps of calculating virtual engine rotary speed according to any one of car speed and motor rotary speed; calculating a combined value of corresponding fundamental frequency and harmonic frequency according to the virtual engine rotary speed; and calculating engine order sound according to the combined value. The electric car active sound generation system and the sound control method thereof provided by the invention have the beneficial effects that a virtual engine rotary speed calculation formula based on electric car speed is provided, and the virtual engine rotary speed of the active sound generation system can be calculated according to the car speed or motor rotary speed, so that a sound frequency produced by the active sound generation system is ensured to range from virtual engine idle speed and rated rotary speed, and authenticity and naturality of a sound produced by the electric car active sound generation system are ensured.

The invention relates to a study to the influence of the rainfall pattern to the ecologic system in the field of ecology, in particular to a device for controlling rainfall shift and an application method thereof. The device comprises a rain gutter, a flow guiding pipe and branch pipes, wherein the rain gutter inclines downwards, a port of the rain gutter is inserted in one end of the flow guiding pipe in a conical shape, a covering part is arranged on the conical part of the rain gutter, the flow guiding pipe is connected with the branch pipes through a three-way connector, the branch pipes are connected to the three-way connector, an end plug is arranged at the tail end of each branch pipe, holes are arranged on the branch pipes, the distances between the corresponding holes on each branch pipe and the three-way connector are identical, and the aperture of each branch pipe is gradually increased from the end where the three-way connector is connected to the tail end. The inclined rain gutter is erected on the ground with previously reduced rainfall, the area of the rain gutter is determined according to the concrete requirement, and the intercepted rainfall flows to a rain trap along the rain gutter and flows to the branch pipes through the flow guiding pipe and then flows to the place with prospective water increasing through the drilling holes on the branch pipes. The invention is scientific and accurate, convenient for implementation and low in cost and meets the requirements of field conditions.

The invention discloses a method and a device for synthesizing emotional speech of a specific speaker under extremely low resources, computer equipment and a storage medium. The method comprises the following steps: acquiring a training text and an audio corresponding to the training text, converting the training text into a phoneme sequence, embedding a slot with an emotion vector to obtain initial training data, inputting into a deep learning model, training to obtain a base model, acquiring a specific text and a specific audio, converting the specific text into a corresponding phoneme sequence, embedding a slot with an emotion vector into the phoneme sequence to obtain specific training data, inputting the specific training data into the base model, training to obtain a speech synthesismodel, converting a to-be-synthesized text into a phoneme sequence to obtain a to-be-synthesized phoneme sequence, filling the to-be-synthesized phoneme sequence into the emotion slot to obtain synthetic input data, and inputting the synthetic input data into the speech synthesis model to obtain a speech audio with specific emotion, so as to reduce the cost of obtaining the speech audio with thespecific emotion and improve the flexibility of an emotion speech synthesis scheme.

The invention belongs to the field of aquatic products and relates to a method improving wheel animalcule growth reproduction capacity and intensifying action of wheel animalcule working as bait nutrient. During cultivation process of aquatic bait wheel animalcules, growth reproduction capacity and an action of wheel animalcule working as bait nutrient can be improved with addition of micro-algae biological products enriched with natural biological active substance. The micro-algae enriched with biological active substance working as pure-natural biological products, aquatic product naturalness, safe use and toxic and side effect-avoidance can be guaranteed via food chain delivery; the micro-algae does not contains various biological active substance, but also is enriched with polysaccharide and protein and other nutrients and substance; and production effect can be comprehensively improved. Anti-oxidization performance, anti-radiation capacity and nutritive effect can be improved by the use of the natural active substance; survival rate, growth speed and anti-disease capacity of seedlings of bred aquatic animal fish, shrimps, crabs and other aquatic animals can be improved; and aquatic culturing efficiency can be comprehensively improved.

The invention provides a semantic-guided dark light image enhancement method, which belongs to the technical field of image processing, and comprises the following steps: S10, converting an original dark light RGB image into an HSV color space image, S20, enhancing the brightness of a V channel in the HSV color space image by using a fusion enhancement model to obtain a V channel after image brightness enhancement, and S30, converting the V channel after image brightness enhancement and the H channel and the S channel in the HSV color space image back to a RGB color space to obtain a brightness enhanced RGB image. The semantic-guided dark light image enhancement method can improve the visibility of image contents in a dark light image so as to reveal details hidden in the dark and can maintain the visual naturalness of the image.

The invention discloses a design method for a virtual reality multi-view fusion model, and the method comprises the following steps: collecting a view type according to an interaction task, obtaining a main view, an auxiliary view, and a configuration mode and a fusion method between the main view and the auxiliary view; and according to the configuration mode and the fusion method, constructing a multi-view fusion model, wherein the multi-view fusion model is used for obtaining a main view image and an auxiliary view image and fusing the auxiliary view image into the main view image to obtain a multi-view fusion image. According to the multi-view fusion model design method, a user determines a configuration mode of a main view angle and an auxiliary view angle and a fusion method of the auxiliary view angle in sequence according to different requirements of an interaction task on user space perception and interaction precision and requirements of the interaction task on the auxiliary view angle in three aspects of information richness, intuition and the user intervention degree; therefore, the appropriate multi-view fusion model is determined, the space information of the surrounding environment is provided for the user, and the correctness of the limb movement posture and the naturalness of interaction are ensured.

The invention discloses a speech synthesis method based on a generative adversarial network. The method comprises the following steps of: S1, preparing training samples, including real audio data, and extracting Mel spectrum features; S2, setting an initialized generator parameter set and a discriminator parameter set; S3, inputting the Mel spectrum features to a generator to obtain an output synthetic audio; S4, correspondingly and simultaneously inputting the real audio data in the S1 and the output synthesized audio obtained in the S3 into a multi-dimensional discriminator and a pooling discriminator; S4, inputting output results of the discriminators into a loss function formula, and respectively calculating a generator loss function and a discriminator loss function; S5, updating the generator and the discriminators; S6, repeating by using the updated generator and discriminators until the set maximum value M of the number of updating times is reached; S7, after each update, returning to the step S3; and S8, performing speech synthesis by using the generator. According to the method, one-dimensional convolution operation is adopted by the generator, model parameters are small, and the speed is high.

The invention discloses a solder strip delivery mechanism of a solar cell serial welding machine and solves the problems in the prior art that easy solder strip breakage and low welding strength are caused by the unreasonable structure of the delivery mechanism. The solder strip delivery mechanism comprises a W-shaped solder strip partition board (16), wherein the W-shaped solder strip partition board (16) is arranged at the bottom of a U-shaped framework (15), strip unwinding rotating shaft bearings (18) and strip unwinding rotating shafts (19) are respectively arranged on two vertical boards of the U-shaped framework (15), a pair of strip unwinding rotating rollers (20) are fixedly arranged on the strip unwinding rotating shafts (19), internally round steps (21) are arranged on the inner sides of the strip unwinding rotating rollers (20), a solder strip reel (23) wound with solder strips (13) is movably clamped between the two internally round steps (21) of the two strip unwinding rotating rollers (20), and reel edges (24) on the two sides of the solder strip reel (23) are connected with rubber rings (22) arranged on the internally round steps (21) in a sleeving way in a roll welding way. The solder strip delivery mechanism has the advantages of convenience in operation and simplicity in structure and is particularly suitable for being applied in the field of solar cell high-speed serial welding.

The invention provides a data fusion technology combined railway linear array color image inpainting algorithm. The algorithm includes the steps: subjecting to-be-inpainted color image to spatial alternation from RGB (red, green and blue) to HIS (hue, saturation and intensity); performing regional characteristic judgment aiming at each pixel in to-be-inpainted regions of sub-images of H, S and I components to divide the regions into flat regions and marginal regions; for pixels in the flat regions, calculating estimation pixel values of the pixels by means of the interpolation algorithm, and using the estimation pixel values to substitute original pixel values of the pixels; for pixels in the marginal regions, calculating estimation gray values of the pixels by means of the interpolation algorithm and a data fusion method, and using the estimation gray values to substitute original gray values of the pixels to obtain inpainted sub-images of the H, S and I components; and converting the inpainted component sub-images from HIS to RGB to obtain a final image. The data fusion technology combined railway linear array color image inpainting algorithm has the advantage that images can be inpainted accurately.

The invention discloses a preparation method of camellia oleosa seed oil for beauty. The preparation method comprises steps as follows: step 1, edible camellia oleosa seed oil is selected from standby application; step 2, camellia seeds are hulled and then are dried to 40%-50% for standby application; step 3, the edible camellia oleosa seed oil in the step 1 is poured into a container, the camellia seeds in the step 2 are put in the container and are soaked for 18-24 h, then residues are fished out, the camellia oleosa seed oil is left to stand for 2-3 h, and a solidified camellia oleosa seed oil product is obtained. The preparation method is simple to operate, and oil can be prepared without squeezing and extraction and can be solidified without addition of any chemicals, so that the product is natural and pollution-free and cannot produce side effects on skin after application.