Abstract: Disclosed is an autonomous distributed system for detecting and tracking a falling object within a region being monitored by using a plurality of camera assemblies cooperating with each other, and disclosed is a camera assembly therefor. Each of the camera assembly, normally, operates in a falling object detecting mode, but, when detecting a falling object, shifts to a tracking mode. In the tracking mode, a degree of danger of the falling object is determined by referring to the size of the falling object in real space and the movement of a falling. In addition, when the falling object is further tracked and moved outside of the angle of view, a search request including information of the degree of danger as to the falling object is issued to surrounding camera assemblies, such that the multiple camera assemblies search for the falling object in cooperation.

Abstract: In a technology for detecting the water level of a river by image processing, after an angle of view setting process, an area setting unit sets an arbitrarily defined certain range from the center of the set angle of view as a processing area. Then, a flow processing unit calculates, from the processing area, motion information and a flow direction, computes a flow density, determines a region having a high density and similar flow directions to be a flow of water currents, and deletes flows in the other directions. Thereafter, a graph-cut processing unit creates an object seed and a background seed for graph-cutting, and detects a water surface by automatic graph-cutting. After an edge extraction processing unit has performed edge extraction, a water level calculation processing unit determines an edge that satisfies a predetermined condition to be a water level line, and outputs a water level measurement result.

Abstract: In a method for determining upper and lower limit values for a target brightness when image contrast is extended, an upper and lower limit value search processing unit establishes two adjacent areas in accordance with brightness of a grayscale histogram, and, while scanning the positions of those areas, compares the frequency of those areas to a threshold, and if one frequency value is greater than or equal to the threshold value and the other frequency value is lower than the threshold, performs upper and lower limit value search processing wherein a brightness value at the boundary of the two areas is determined as an upper or lower limit value. Thresholds for upper and lower limit value search start position and frequency are established based on the shape of the grayscale histogram of an image to be processed. The shape of the grayscale histogram is identified according to preset classifications.

Abstract: In a technology for detecting the water level of a river by image processing, after an angle of view setting process, an area setting unit sets an arbitrarily defined certain range from the center of the set angle of view as a processing area. Then, a flow processing unit calculates, from the processing area, motion information and a flow direction, computes a flow density, determines a region having a high density and similar flow directions to be a flow of water currents, and deletes flows in the other directions. Thereafter, a graph-cut processing unit creates an object seed and a background seed for graph-cutting, and detects a water surface by automatic graph-cutting. After an edge extraction processing unit has performed edge extraction, a water level calculation processing unit determines an edge that satisfies a predetermined condition to be a water level line, and outputs a water level measurement result.

Abstract: In a method for determining upper and lower limit values for a target brightness when image contrast is extended, an upper and lower limit value search processing unit establishes two adjacent areas in accordance with brightness of a grayscale histogram, and, while scanning the positions of those areas, compares the frequency of those areas to a threshold, and if one frequency value is greater than or equal to the threshold value and the other frequency value is lower than the threshold, performs upper and lower limit value search processing wherein a brightness value at the boundary of the two areas is determined as an upper or lower limit value. Thresholds for upper and lower limit value search start position and frequency are established based on the shape of the grayscale histogram of an image to be processed. The shape of the grayscale histogram is identified according to preset classifications.

Abstract: Disclosed is an autonomous distributed system for detecting and tracking a falling object within a region being monitored by using a plurality of camera assemblies cooperating with each other, and disclosed is a camera assembly therefor. Each of the camera assembly, normally, operates in a falling object detecting mode, but, when detecting a falling object, shifts to a tracking mode. In the tracking mode, a degree of danger of the falling object is determined by referring to the size of the falling object in real space and the movement of a falling. In addition, when the falling object is further tracked and moved outside of the angle of view, a search request including information of the degree of danger as to the falling object is issued to surrounding camera assemblies, such that the multiple camera assemblies search for the falling object in cooperation.