Abstract: A training method for video stabilization and an image processing device using the same are proposed. The method includes the following steps. An input video including low dynamic range (LDR) images is received. The LDR images are converted to high dynamic range (HDR) images by using a first neural network. A feature extraction process is performed to obtain features based on the LDR images and the HDR images. A second neural network for video stabilization is trained according to the LDR images and the HDR images based on a loss function by minimizing a loss value of the loss function to generate stabilized HDR images in a time-dependent manner, where the loss value of the loss function depends upon the features. An HDR classifier is constructed according to the LDR images and the HDR images. The stabilized HDR images are classified by using the HDR classifier to generate a reward value, where the loss value of the loss function further depends upon the reward value.

Abstract: A training method for video stabilization and an image processing device using the same are proposed. The method includes the following steps. An input video including low dynamic range (LDR) images is received. The LDR images are converted to high dynamic range (HDR) images by using a first neural network. A feature extraction process is performed to obtain features based on the LDR images and the HDR images. A second neural network for video stabilization is trained according to the LDR images and the HDR images based on a loss function by minimizing a loss value of the loss function to generate stabilized HDR images in a time-dependent manner, where the loss value of the loss function depends upon the features. An HDR classifier is constructed according to the LDR images and the HDR images. The stabilized HDR images are classified by using the HDR classifier to generate a reward value, where the loss value of the loss function further depends upon the reward value.

Abstract: A training method for video stabilization and an image processing device using the same are proposed. The method includes the following steps. An input video including low dynamic range (LDR) images is received. The LDR images are converted to high dynamic range (HDR) images by using a first neural network. A second neural network for video stabilization is trained to generate stabilized HDR images in a time-dependent manner.

Abstract: Provided are liners for a shaped charge and corresponding methods of use. An example liner comprises a generally conical shape having an apex, an open side, a liner wall comprising a thickness, and an axis extending through the center of the liner from the apex to the center of the open side. The liner comprises a liner height extending in a vertical plane from the center of the open side to the apex, a liner radius extending along a horizontal plane that is perpendicular to the axis at the open side of the liner and that extends from the axis to an outermost edge of the liner wall. The ratio of the liner height to the liner diameter is about 0.90 to about 1.10. The liner wall comprises an apex curvature, a first wall curvature, a second wall curvature, and a third wall curvature.

Abstract: A training method for video stabilization and an image processing device using the same are proposed. The method includes the following steps. An input video including low dynamic range (LDR) images is received. The LDR images are converted to high dynamic range (HDR) images by using a first neural network. A second neural network for video stabilization is trained to generate stabilized HDR images in a time-dependent manner.

Abstract: Provided are liners for a shaped charge and corresponding methods of use. An example liner comprises a generally conical shape having an apex, an open side, a liner wall comprising a thickness, and an axis extending through the center of the liner from the apex to the center of the open side. The liner comprises a liner height extending in a vertical plane from the center of the open side to the apex, a liner radius extending along a horizontal plane that is perpendicular to the axis at the open side of the liner and that extends from the axis to an outermost edge of the liner wall. The ratio of the liner height to the liner diameter is about 0.90 to about 1.10. The liner wall comprises an apex curvature, a first wall curvature, a second wall curvature, and a third wall curvature.

Abstract: A multi-checkpoint type clustered animal counting device is proposed, which is capable of providing a counting function that can be used for statistically determining the number of animals (such as fruit flies) within a region such as farmland or garden. The proposed animal counting device is characterized by the utilized to at least two object sensors, wherein the first object sensor is disposed at a first checkpoint while the second object sensor is disposed at a second checkpoint, and wherein the first object sensor is initially set to power-on state while the second object sensor is initially set to power-off state and can be switched on only when the first object sensor is triggered. When the second object sensor is triggered, the counting operation will increase the output count number by one. This feature allows a more accurate result and can help save power consumption.

Abstract: A forward and backward image resizing method is used for resizing a low-resolution image into a high-resolution image. In the method, the low-resolution image is obtained first, and then a forward and backward image resizing process is performed, so as to resize the low-resolution image into the high-resolution image with an integral multiple resolution. The forward and backward image resizing process includes: resizing the low-resolution image by the integral multiple, so as to generate a first-resizing image with the integral multiple resolution; further increasing the resolution of the first-resizing image by 2-fold, so as to generate a second-resizing image; and reducing the resolution of the second-resizing image by 2-fold, thereby obtaining the high-resolution image.

Abstract: A front-end gateway unit is designed for integration to a remote ecological environment monitoring system that is equipped with a wireless sensor network (WSN) system installed at a remote site, such as a farmland or a garden, for the purpose of allowing the WSN system to exchange data with a back-end host server via a wireless communication system. The front-end gateway unit is characterized by the capability of using either the WSN system or a built-in sensing module for collecting ecological data, and the capability of combining geographical location data in the ecological data. This feature allows the collection of a comprehensive set of ecological data (including geographical location, temperature, humidity, sunlight data, wind speed, and pest number) for transfer to the back-end host server, such that research/management personnel at the local site can conveniently browse these ecological data and learn the ecological conditions of the remotely monitored area.

Abstract: A wireless-linked remote ecological environment monitoring system is proposed, which is characterized by the use of a sensor network such as WSN (wireless sensor network) installed at the remote site for collecting ecological data, and the use of a public wireless communication system such as GSM (Global System for Mobile Communications) for transferring all the collected ecological data to a back-end host server unit where the ecological data are compiled into webpages for posting on a website. This feature allows the research/management personnel to browse the ecological data simply by linking a network workstation via a network system such as the Internet to the website, without having to travel to the remote site and collect ecological data by human labor.

Abstract: A forward and backward image resizing method is used for resizing a low-resolution image into a high-resolution image. In the method, the low-resolution image is obtained first, and then a forward and backward image resizing process is performed, so as to resize the low-resolution image into the high-resolution image with an integral multiple resolution. The forward and backward image resizing process includes: resizing the low-resolution image by the integral multiple, so as to generate a first-resizing image with the integral multiple resolution; further increasing the resolution of the first-resizing image by 2-fold, so as to generate a second-resizing image; and reducing the resolution of the second-resizing image by 2-fold, thereby obtaining the high-resolution image.

Abstract: A junction-free NAND flash memory is described, including a substrate, memory cells, source/drain inducing (SDI) gates electrically connected with each other, and a dielectric material layer. The memory cells are disposed on the substrate, wherein each memory cell includes a charge storage layer. Each SDI gate is disposed between two neighboring memory cells. The dielectric material layer is disposed between the memory cells and the SDI gates and between the SDI gates and the substrate.

Abstract: A junction-free NAND flash memory is described, including a substrate, memory cells, source/drain inducing (SDI) gates electrically connected with each other, and a dielectric material layer. The memory cells are disposed on the substrate, wherein each memory cell includes a charge storage layer. Each SDI gate is disposed between two neighboring memory cells. The dielectric material layer is disposed between the memory cells and the SDI gates and between the SDI gates and the substrate.

Abstract: A wireless sensor network gateway unit is proposed, which is designed for integration to a wireless sensor network (WSN) for providing a gateway function with a failed link auto-redirecting capability for the wireless sensor network. The proposed WSN gateway unit is characterized by the provision of an failed link auto-redirecting capability, which can respond to the failure of any sensor node in the WSN system by performing a failed link auto-redirecting operation for redirecting the down-linked good sensor nodes for linking to a nearby good sensor node to thereby allow the down-linked good sensor nodes to be nevertheless able to transfer data to the WSN gateway unit of the invention. This feature allows the WSN gateway unit of the invention to maintain good operational reliability for the WSN system.

Abstract: A wireless sensor network gateway unit is proposed, which is designed for integration to a wireless sensor network (WSN) for providing a gateway function with a failed link auto-redirecting capability for the wireless sensor network. The proposed WSN gateway unit is characterized by the provision of an failed link auto-redirecting capability, which can respond to the failure of any sensor node in the WSN system by performing a failed link auto-redirecting operation for redirecting the down-linked good sensor nodes for linking to a nearby good sensor node to thereby allow the down-linked good sensor nodes to be nevertheless able to transfer data to the WSN gateway unit of the invention. This feature allows the WSN gateway unit of the invention to maintain good operational reliability for the WSN system.

Abstract: A back-end host server unit for remote ecological environment monitoring system is designed to provide a back-end host server function for a front-end gateway unit and sensor network, such as WSN (Wireless Sensor Network) installed at a remote site such as farmland. This back-end host server unit is characterized by the capability of using a public wireless communication system, such as GSM (Global System for Mobile Communications), for receiving ecological data from the front-end gateway unit and WSN system, and the capability of compiling received ecological data into webpages for posting on a website that allows the research/management personnel to browse the ecological data via a network system such as the Internet. This feature allows the research/management personnel at the local site to conveniently gather ecological data and learn the ecological conditions of the remote site.

Abstract: A front-end gateway unit is designed for integration to a remote ecological environment monitoring system that is equipped with a wireless sensor network (WSN) system installed at a remote site, such as a farmland or a garden, for the purpose of allowing the WSN system to exchange data with a back-end host server via a wireless communication system. The front-end gateway unit is characterized by the capability of using either the WSN system or a built-in sensing module for collecting ecological data, and the capability of combining geographical location data in the ecological data. This feature allows the collection of a comprehensive set of ecological data (including geographical location, temperature, humidity, sunlight data, wind speed, and pest number) for transfer to the back-end host server, such that research/management personnel at the local site can conveniently browse these ecological data and learn the ecological conditions of the remotely monitored area.

Abstract: A wireless-linked remote ecological environment monitoring system is proposed, which is characterized by the use of a sensor network such as WSN (wireless sensor network) installed at the remote site for collecting ecological data, and the use of a public wireless communication system such as GSM (Global System for Mobile Communications) for transferring all the collected ecological data to a back-end host server unit where the ecological data are compiled into webpages for posting on a website. This feature allows the research/management personnel to browse the ecological data simply by linking a network workstation via a network system such as the Internet to the website, without having to travel to the remote site and collect ecological data by human labor.

Abstract: A multi-checkpoint type clustered animal counting device is proposed, which is capable of providing a counting function that can be used for statistically determining the number of animals (such as fruit flies) within a region such as farmland or garden. The proposed animal counting device is characterized by the utilized to at least two object sensors, wherein the first object sensor is disposed at a first checkpoint while the second object sensor is disposed at a second checkpoint, and wherein the first object sensor is initially set to power-on state while the second object sensor is initially set to power-off state and can be switched on only when the first object sensor is triggered. When the second object sensor is triggered, the counting operation will increase the output count number by one. This feature allows a more accurate result and can help save power consumption.