Abstract: An object detection device acquires a plurality of images captured by a camera mounted in a moving body and acquires movement information of the camera. The object detection device extracts an edge from a first image among the acquired images. The object detection device calculates a plurality of candidate points corresponding to the edge in a second image, based on the movement information and calculates candidate points corresponding to a plurality of heights as height candidates of the edge. The object detection device calculates a correlation index value between the edge and each of the plurality of candidate points and detects the candidate point with the largest correlation as a correspondence point of the edge. The object detection device detects an object at a position of the edge when the correlation index value of the detected correspondence point is equal to or smaller than a predetermined threshold.

Abstract: An object detection device acquires a plurality of images captured by a camera mounted in a moving body and acquires movement information of the camera. The object detection device extracts an edge from a first image among the acquired images. The object detection device calculates a plurality of candidate points corresponding to the edge in a second image, based on the movement information and calculates candidate points corresponding to a plurality of heights as height candidates of the edge. The object detection device calculates a correlation index value between the edge and each of the plurality of candidate points and detects the candidate point with the largest correlation as a correspondence point of the edge. The object detection device detects an object at a position of the edge when the correlation index value of the detected correspondence point is equal to or smaller than a predetermined threshold.

Abstract: A FMCW-type radar device generates snapshot data from a beat signal that represents a received condition of the radar device every modulation period. Auto-correlation matrices generated by the snapshot data every modulation period are averaged every set of plural periods. The radar device calculates the target azimuth of a target object such as a preceding vehicle based on the averaged auto-correlation matrix based on MUSIC (MUltiple SIgnal Classification) method. This averaging is performed by weighting average based on an amount of mixed noise (or an interference amount) contained in the snapshot data in each modulation period. A weighting coefficient to be applied to the auto-correlation matrix in each modulation period is set to a value corresponding to the amount of mixed noise, namely, the interference amount of this modulation period. The weighting coefficient becomes large when the interference amount is small, and on the other hand, becomes small when it is large.

Abstract: A technology for providing an arrival direction estimation apparatus which can greatly reduce the calculation amount in spectrum calculation and can perform precise direction estimation without setting short frequency division in spectrum calculation is disclosed.

Abstract: In a FM-CW or CW radar apparatus, when interference components are contained in channel signals obtained as beat signals from array antenna elements of respective channels, and the interference components result from directly receiving transmitted CW radar waves from an external source, phase shifting is applied to each of the channel signals to shift respective phases of the interference components of the respective channels to a condition corresponding to reception of interference waves from a predetermined direction. The interference components are then eliminated, and reverse phase shifting is applied to restore remaining components of the channel signals to their original phase condition.

Abstract: A FMCW-type radar device generates snapshot data from a beat signal that represents a received condition of the radar device every modulation period. Auto-correlation matrices generated by the snapshot data every modulation period are averaged every set of plural periods. The radar device calculates the target azimuth of a target object such as a preceding vehicle based on the averaged auto-correlation matrix based on MUSIC (MUltiple SIgnal Classification) method. This averaging is performed by weighting average based on an amount of mixed noise (or an interference amount) contained in the snapshot data in each modulation period. A weighting coefficient to be applied to the auto-correlation matrix in each modulation period is set to a value corresponding to the amount of mixed noise, namely, the interference amount of this modulation period. The weighting coefficient becomes large when the interference amount is small, and on the other hand, becomes small when it is large.

Abstract: In a FM-CW or CW radar apparatus, when interference components are Contained in channel signals obtained as beat signals from array antenna elements of respective channels, and the interference components result from directly receiving transmitted CW radar waves from an external source, phase shifting is applied to each of the channel signals to shift respective phases of the interference components of the respective channels to a condition corresponding to reception of interference waves from a predetermined direction. The interference components are then eliminated, and reverse phase shifting is applied to restore remaining components of the channel signals to their original phase condition.

Abstract: Apparatus and method are disclosed for estimating direction of an arrival wave, such as radio or acoustic waves. A correlation matrix of an arrival wave is computed, and matrices for obtaining eigenvalues and eigenvectors are produced. Column norms of the matrix, which is subject to a Householder transformation, are computed. Whether a maximum value of the column norms is the same as or lower than a predetermined threshold is judged. A Householder transformation of the matrix is performed only if the maximum value of the column norms is determined to be greater than or equal to the threshold, and not when the maximum value of the column norms is less than the threshold. The eigenvectors and the eigenvalues of the correlation matrix are obtained by further implementing the Householder transformation. The number of arrival waves is judged from the number of times the Householder transformation is performed.

Abstract: A technology for providing an arrival direction estimation apparatus which can greatly reduce the calculation amount in spectrum calculation and can perform precise direction estimation without setting short frequency division in spectrum calculation is disclosed.

Abstract: Apparatus has means for computing correlation matrix of arrival wave, means for producing a matrix for obtaining eigenvalues and a matrix for obtaining eigenvectors, means for computing column norms on the matrix which is a subject for Householder transformation, means for judging whether a maximum value of the column norms is the same as or lower than a predetermined threshold, means for repeating Householder transformation on the matrix only if the maximum value of the column norms is judged not to be the same or not to be lower, means for breaking Householder transformation if the maximum value is the same as or lower, means for further implementing Householder transformation and for obtaining the eigenvectors and the eigenvalues of the correlation matrix, and means for judging the number of the arrival waves from the implementation times of Householder transformation.