Showing papers on "Centroid published in 1986"

Dual mode video tracker

Abstract: Point and area target tracking are employed by a dual mode video tracker which includes both a correlation processor (30) and a centroid processor (32) for processing incoming video signals representing the target scene (20) and for generating tracking error signals over an entire video frame. A microcomputer (26) controls the operation of the correlation and centroid processors, as well as a preprocessor (28) which preconditions the incoming video signals. The tracker operates in any of several modes in which the correlation and centroid processors function either independently or cooperatively with each other. The correlation processor generates a reference map in pixel format which is used to produce azimuth and elevation optimal weighting values for each pixel in the field-of-view. The reference map is recursively updated. The track gate for the correlation processor is automatically sized using a gradient function derived from the optimal weighting values. The gradient function is also employed along with difference video to determine correlation tracking status. The correlation and centroid tracking errors may be combined to form a composite signal defining a true complementary function.

Optimal Doppler centroid estimation for SAR data from a quasi-homogeneous source

Abstract: In synthetic aperture radar (SAR) signal processing, an accurate Doppler centroid is required for most applications involving target motion estimation and antenna pointing direction estimation. In some cases the Doppler centroid can be sufficiently determined using available information regarding the terrain topography, the relative motion between the sensor and the terrain, and the antenna pointing direction. But most often, a highly accurate Doppler centroid value has to be derived by analyzing the received SAR signal itself. This kind of signal processing is referred to as Doppler centroid estimation (DCE). This correspondence briefly describes two DCE algorithms, provides a performance summary for these algorithms, and presents the experimental results. These algorithms include a previously reported one and a newly developed one that is optimized for quasi-homogeneous sources. The performance enhancement achieved by the optimal DCE algorithm is clearly demonstrated by the experimental results.

Sensor fusion and object localization

Abstract: In this paper, we discuss the issue of locating objects through multiple sensory information Sensor measurements are subject to limitations of sensor precision and accuracy Although errors in position estimates are affected only by the errors of sensor measurements, errors in orientation estimates are also dependent on the dimensions over which the measurement has been made The concept of good measurement is used in selecting and weighting partial estimates of the position and orientaion The problem of finding the best estimate of the position and orientation is formulated as a linear system of these multiple estimates The best estimate is then obtained by solving this system in a weighted least square sense This method has been implemented for a manipulator end-effector instrumented with centroid and matrix tactile sensors

Practical land grading based on least squares

Abstract: Practical methods for land grading design of a plane surface for rectangular and irregularly shaped fields based on a least squares analysis are presented. The least squares procedure leads to a system of three linear equations with three unknowns for determination of the best‐fit plane. The equations can be solved by determinants (Cramer's rule) using a procedure capable of solution by many programmable calculators. The detailed computational process for determining the equation of the plane and a simple method to find the centroid location of an irregular field are also given. An illustrative example and design instructions are included to demonstrate the application of the design procedure.

Centroid Mapping Optimization: A New Efficient Optimization for Food Research and Processing

Abstract: Experimenting at the centroid during simplex optimization was efficient in search for the optimum. A centroid search process, which replaced the worst vertex within the simplex with the centroid vertex one after another, was introduced. In the new method, the optimum can be found by combination of initial simplex, centroid search, mapping and simultaneous factor shift. An algorithm to define the new range for the subsequent simplex was also applied when the centroid search did not work effectively. This new method enabled efficient research by conducting a set of experiments simultaneously. By changing the range in sequential simplex/centroid search, the optimum could be more efficiently found than by previous simplex methods when compared with mathematical models.

Variance expressions for four spectral-centroid estimators of a complex Gaussian process

Abstract: Variance expressions are derived for four unbiased centroid estimators of an asymmetrical spectrum of the complex envelope of a narrow-band Gaussian process. These estimators are obtained by cross-correlating the undistorted or hard-clipped component and the time derivative of the undistorted or hard-dipped quadrature component of the complex envelope. When both components are hard-clipped, the variance expression of the centroid estimator is a generalization of Blachman's zero-crossing analysis. The variance of the centroid estimator obtained from finite records of the undistorted components is shown to be smaller than that of the other three estimators when m_{2} > 5.32\mu_{1}^{2} , where \mu_{1} and m_{2} are the first simple and second central spectral moments respectively. However, when m_{2} , the variance of the centroid estimate obtained from one hard-clipped component is shown to be smallest. Furthermore, the dependence of the variances on both the magnitude and direction of spectral asymmetry is shown to be different for the four centroid estimators considered.

Picture displaying method

Abstract: PURPOSE:To simplify the interpolation calculation by determining the pixel on 2-dimensional display scope to which triangular area cut out from 3 dimensional object it belongs using the centroid of the triangular area and 3 side members of the triangle. CONSTITUTION:The surface of object is cut out into triangular unit area, and the data composed of object surface information connecting the 3 apexes of the triangular unit area. Out of these data, the position data of the apexes are perspectively converted into 2 dimensional plane which corresponds to display screen according to the sight position, and the data is obtained to indicate apex position on 2 dimensional plane which correspond with the 3 apexes. The centroid of the triangular area are obtained from these data, position relation between the centroid and pixel is judged, and the pixel is judged whether it is in the triangular area. Then, the object furnace information data is vertically placed at the apex position of the triangular area, and the triangular data plane is placed on the point edge of its, and interpolation calculation is made for the triangular area.

Arrangement judging device

Abstract: PURPOSE:To check the validity of arrangement of parts as a judgement object in a picture and check the alignment of parts also in case of plural ojbects by providing a means, which calculates the centroid of a shape as a judgement object, and a means which calculates the direction. CONSTITUTION:A conventional shape judging device is provided with a centroid calculating circuit 311 and a direction calculating circuit 312 additionally. The circuit 311 calculates the centroid of each independent shape separated by the connectivity processing in a connectivity processing circuit 103. The direction calculating circuit 312 calculates the rotation direction of the shape. In case of check, the two-dimensional shape of parts 100 to be checked with a base is picked up by a two-dimensional image pickup device 101, and arrangement feature absolute values and arrangement feature relative values obtained through an A/D converter 102 and circuits 311 and 312 are evaluated. As the result, the correct arrangement of parts on the base is judged even if parts are placed and rotated freely on the base picture. Further, the validity of the arrangement of parts as the judgment object in the picture is checked, and the alignment of parts is checked also in case of plural objects.

High accuracy image centroid position determination with matrix sensors: An experimental comparison of methods

Abstract: There is increasing utilization of optical position measurement techniques in research and industry as demands for precision increase. A number of methods have been developed for this purpose. In the present study attention was directed toward the use of matrix array sensors such as the CCD or CID sensors used in digital cameras. The primary objective of the study was to determine the accuracy with which the centroid of a laser beam or the image of a circular incoherently illuminated object (such as a mirror) could be determined. Five different methods for finding the image centroid were compared experimentally to determine which would be preferable. These are the intensity, Kortegaard, quadrant, Fourier descriptor, and cross-correlation centroid methods. The quadrant centroid method is presented here for the first time and the other methods are briefly reviewed. Experimental results show that the intensity and cross-correlation methods had the best accuracies for both the laser beam and incoherent circular images. In both cases the maximum error observed was about 1.7 μm in-the image plane using a sensor with a pixel spacing of approximately 46 μm. The effects of laser beam wander and beam wander compensation methods were also studied. It was found that beam wander due to the thermal boundary layer at the front face of a laser can be largely eliminated by attaching an open extension tube or enclosed insulated tunnel to the laser.There is increasing utilization of optical position measurement techniques in research and industry as demands for precision increase. A number of methods have been developed for this purpose. In the present study attention was directed toward the use of matrix array sensors such as the CCD or CID sensors used in digital cameras. The primary objective of the study was to determine the accuracy with which the centroid of a laser beam or the image of a circular incoherently illuminated object (such as a mirror) could be determined. Five different methods for finding the image centroid were compared experimentally to determine which would be preferable. These are the intensity, Kortegaard, quadrant, Fourier descriptor, and cross-correlation centroid methods. The quadrant centroid method is presented here for the first time and the other methods are briefly reviewed. Experimental results show that the intensity and cross-correlation methods had the best accuracies for both the laser beam and incoherent circul...

Instrument for measuring deviation value of centroid position in radius direction of rotary center axis of cylindrical body

Abstract: PURPOSE:To measure the deviation value of a centroid position in the radius direction of the rotary center axis of a cylindrical body by finding out the coordinates of the rotary centroid of the cylindrical body. CONSTITUTION:Driving chucks 2, 3 are fitted to both the ends of the thin cylindrical body 1 and supported by two rollers each. The center of the chuck 2 is coupled with a driving device 6 through a universal joint 4, an angle detector 7 for measuring the turning angle of the cylindrical body 1 is fitted to the device and a thickness measuring sensor 9 is fitted to the external surface of the cylindrical body 1. The deviation delta and thickness (t) of the external size between a reference circle 12 and the cylindrical body 1 which corresponds to the positional angle theta measured by the angle detector 7 are measured by a deviation measuring device 8 having a sensor contacted with the external surface of the cylindrical body 1 and to be reciprocated in accordance with the deviation of the centroid position based upon the shape error and positional shift of the cylindrical body 1 at respective measuring point (i) obtained by dividing the outer periphery of the cylindrical body 1 into n parts and the coordinates Zo, Yo of the centroid G to the center B of the reference circle 12 are calculated. Consequently, the deviation of the radius direction from the rotary center shaft at an optional position in a longitudinal direction can be found out.

Method for teaching specific position of target object in visual sensor system

Abstract: PURPOSE:To teach the specific position of the target object in a visual sensor system by teaching said position by the coordinate value of the object coordinate system XY of which the origine is the controid of the plane image of the target object and the main axis is the X-axis or Y-axis. CONSTITUTION:A microcomputer 14 calculates the characteristic value of a prescribed characteristic term and collates the same with the reference characteristic value stored in a RAM18 thereby discriminating the kind of the object of this time. The microcomputer determines the centroid and main axis of the target object, presumes the object coordinate system and calculates the coordinate value in the coordinate system of the visual sensor in the taught specific position from the coordinate value in the object coordinate system of the specific position stored in the RAM18. The microcomputer informs a robot of the information on the kind of the target object, the coordinate value in the visual sensor coordinate system of the specific position P, the position of the centroid and the attitude angle thetaob of the target object via an interface circuit 13. The teaching is thus executed simply and easily as the position of the target object is taught by the coordinate value of the X-Y coordinates.

Method and device for character reading of printed label

Abstract: PURPOSE:To recognize characters accurately and at a high speed by obtaining the segment center coordinates through an interior division calculation and based on the data on an interior division ratio of positions of characters stored previously after detecting the four corners and the centroid coordinates of a black frame and then discriminating characters by a segment window and in response to said segment center coordinates. CONSTITUTION:An end point detector 16 detects the coordinates of four corners P1-P4 of a black frame 12 in a processed picture 7. While a centroid detector 17 detects the coordinates of the centroid (g) of the frame 12. Both detectors 16 and 17 supply the data on the processed picture via a frame detector 15 and deliver the data on the detected corner and centroid coordinates to a segment position detector 18. A table 19 storing the data on an interior division ratio of characters 3 on a lable 1 is connected to the detector 18. The detector 18 detects the segment center coordinates based on said data on the interior division ratio and on the centroid coordinates as well as the corner coordinates. A character deciding circuit 20 sets a segment window in accordance with the segment center coordinates and decides the presence/absence of segments set under the segment window for recognition of characters.

Label attachment device to hole

Abstract: PURPOSE:To omit an arraying process of parts by deciding a hole serving as a base point after comparison of feature quantities of each hole and attaching labels to each hole toward a prescribed revolving direction from a hole centroid serving as a base point centering on the centroid of a parts and along the centroid of each hole. CONSTITUTION:A parts 100 to be decided is image-picked up by a 2-dimensional image pickup device 102 and binary coded 302 by the prescribed threshold value to obtain a parts picture where the parts is separated from its background. Then the coordinates of the centroid G0 of the parts are obtained together with other feature quantities of the parts picture. At the same time, the holes H11-H13 existing in the parts picture are detected. Based on these detection data, the areas S1-S3 of each hole are calculated. Then the surrounding lengths l1-l3 of each hole are obtained together with the coordinates of centroids G1-G3 of each hole. Then distances L1-L3 are obtained among centroids are obtained together with angles theta1-theta3 among adjacent segments. A position/ rotation normalizing circuit 309 controls the parts 100 so that the labels are attached to the same hole regardless of the rotation or position variance of the parts 100 based on the above-mentioned feature quantities.

Method for determining reference point position of picture

Abstract: PURPOSE:To accurately settle the position of a reference point, by dividing the picture elements of outlines and finding the proportion of a part in an area and the position of a geometric feature point, such as the center of gravity, etc., by adding the part to the area. CONSTITUTION:A picture element string extracting circuit 3 extracts the picture element strings of a designated area from a picture image memory 2 and a painting-out circuit 4 paints out the extracted picture element strings. A picture element dividing circuit 5 reads out the luminance of the painted-out picture elements and the picture elements in the vicinity of the painted-out ones and divides the area into two, an inside area and outside area, and then, finds a reference area. An accumulated luminance value counting circuit 6 successively counts the luminance value in the reference area by regarding them as uniform and a centroid position determing circuit 7 determines the centroid position by using an accumulated luminance graph implemented by the circuit 6 and finds the position of a specified point (GCP) on a map from the position of the center of gravity, and then, outputs the position.

Method for detecting center of incomplete circle

Abstract: PURPOSE: To improve center detecting accuracy more than the resolution of a pixel and to find out a complete center by arranging windows on an incomplete circle to be a non-detecting object so as to exclude an incomplete circle part to fetch a partial image and finding out the centroid position. CONSTITUTION: A camera 5 for an image processor 6 is fixed and the objective incomplete circle 3 is moved every fixed values ΔX, ΔY by an X-Y table 4 so that respective windows 1, 2 are arranged on the incomplete circle 3. A rectangular window 1 whose longitudinal direction is parallel with the X axis is set up so as to exclude an incomplete circle part (spline), its image is fetched to the device 6 and the coordinates X 1 , Y 1 of he centroid P 1 of its inner circle part are found out. Then, a rectangular window 2 wholse longitudinal direction is parallel with the Y axis is similarly set up so as to exclude the incomplete circle part, its image is fetched to the processor 6 and the coordinates of the point P 1 in the inner circular part are found out. The center detecting accuracy is improved more then the resolution of the pixel and the center of an incomplete circle such as a gear can be highly accurately detected. COPYRIGHT: (C)1987,JPO&Japio

Asymmetry in some common assignment algorithms: the dispersion factor solution

Abstract: Many common assignment algorithms are based on Dial's original design to determine the paths that trip makers will follow from a given origin to destination centroids. The purpose of this paper is to show that the rules that have to be applied result in two unwanted properties. The first is that trips assigned from an origin centroid i to a destination j can be dramatically different to those resulting from centroid j to centroid i, even if the number of trips is the same and the network is perfectly symmetric. The second is that the number of paths that are ‘accepted’ by the algorithm can be surprisingly small. Asymmetry can cause numerous problems in transport model applications. It will not only produce unrealistic results, but also, as is demonstrated here, makes convergence toward supply — demand equilibrium impossible. The paper describes the principles on which Dial-type algorithms are based, the way in which they affect the transport model structure, and how the unwanted characteristics arise. An ...

Centroid point operator

Abstract: PURPOSE:To calculate a centroid point without performing the multiplication processing by reading the binary picture data out of a memory to calculate the addition value of the column and row centroids from the contents of the binary picture data and then dividing said addition value by the number of detected picture elements. CONSTITUTION:Both counters 2 and 4 count the clock pulses and send the count values 3 and 5 to a memory 1 for designation of the columns and rows of the memory 1. At the same time, these columns and rows are sent to the columns and row centroid adders 11 and 12 respectively. The binary picture data 8 read out of the memory 1 are sent to a detected picture element counter 9 and adders 11 and 13. Here the counter 9 counts up pictures if included in the data 8. While adders 11 and 13 add the count values 3 and 5 to the column and row addition values 12 and 14 respectively and produce new addition values 12 and 14 to send them to a divider 15. Thus the divider 15 devides both values 12 and 14 by the detected picture element number 10 respectively and obtain the column and row centroid coordinates. Thus the multiplication processing can be omitted for operation of the centroid points.

Work detecting method

Abstract: PURPOSE:To shorten the analysis time of data by comparing signal work data based upon the brightness difference of an image signal on a circumference set having its center at the centroid position on the drawing of a work with reference data and finding the correlativity of the data. CONSTITUTION:The shape of the work 13 is read by a visual sensor 14 and its image is converted into a binary signal of '1' and '0' to compute the position of the centroid P of the drawing of the work 13 from the image signal. Plural virtual circles (j)... are formed which have their centers at the centroid P and pass through feature parts of the work 13 and image parts overlapping with the respective circles (j) are divided by a constant angle; and the brightness of an image at each division position is converted into a binary signal of '1' and '0' and binary-coded signals of j=1-m on the set circumferences are used as word data Dj(theta) to calculate the correlativity with preset reference data Dij(theta), thereby detecting the position, attitude, and kind of the work 13.

and Tracking of Moving Images Using Adaptive Three-Dimensional Recursive Filters

Abstract: A three-dimensional (3-D) adaptive filter control system is described which is capable of seeking, tracking and enhancing an object that moves with time on an arbitrary, but smooth, trajectory in a dynamic digital image. Enkancement of the object is performed by employing a time-varying first-order 3-D planar-resonant recursive filter that is adapted via its coefficients at each time interval to correspond to the trajectory of the object. The trajectory of the object is determined at each time interval by tracking the intensity centroid of the object at the output of the adaptive 3-D recursive filter. The system is initialized for tracking by a seek algorithm which searches the image for the first candidate object having uniform motion and intensity. Experimental evidence is provided showing the effectiveness of the system in locating, tracking and enhancing objects in images containing other objects and high levels of noise.

Recognition system for on-line handwritten character

Abstract: PURPOSE:To simplify a character recognition system by approximating each stroke at many points for characters like HIRAGANA (Japanese syllabary) including curve components and securing the resemblance with a standard character approximated at a point where the strokes are thinned among those standard characters approximat ed at the same point. CONSTITUTION:The input character sent from a character information input part undergoes the conversion of coordinates so that the centroid point is coincident with an original point as well as the normalization so as to secure a fixed distance between the centroid point and each stroke point. Thus the number 201 of strokes is detected. Each stroke is approximated at a point 2N-1 including both start and end points at an approximate point extracting part 3. Then an inter-pattern distance calculating part 4 obtains the distance between approximate points for a thinned point N and a pattern stored in a standard pattern memory part 7. For the result of this calculation, a minimum distance detecting part 5 detects the minimum and the next minimum values. Then the part 4 obtains a ratio between both values and delivers a code corresponding to a category of the stroke showing the minimum value when said ratio is larger than a certain threshold value through a switch part 10. Then the correction is carried out based on a category code for discrimination from an approximate character via a resemblance table 8.

Centroid tracking with area array detectors

Abstract: A computer program (ALGEVAL) has been developed to simulate the position estimating behavior of a centroid estimator algorithm using data typical of optical point spread function data recorded by an area array detector. Typical results are shown of varying detector properties and optical point spread function types. The detector parameters currently available for study include read noise mean value, dark current mean value and spatial variation, charge transfer efficiency and point spread function location, saturation level, signal level and pixel size. The program is capable of calculating any order centroid using an array size from 2 x 2 to 15 x 15 pixels. The output of the program is either a performance map, histogram data or tabluar data. A number of further developments are recommended.