JP2008293172A - Disappearing dot estimation device and program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To make it possible to stably estimate a position of a disappearing dot even when a plurality of straight lines are not detected. <P>SOLUTION: Predicted hypothetical dots in a time t are generated based on existence level distribution generated based on an image picked up last time (106), and the pickup image picked up in the time t is acquired from an imaging device 12 (108), and a plurality of featured points are extracted from the pickup image (110), and at least one straight line is detected from the pickup image (112). Then, weight at each of the predicted hypothetical dots is updated, and existence level distribution is generated based on a distance between the detected straight line and each of the predicted hypothetical dots (114), and the predicted hypothetical dot as the maximum weight is retrieved from the existence level distribution, and the position of the retrieved predicted hypothetical dot is estimated as the position of the disappearing dot on the pickup image (116). <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to a vanishing point estimation apparatus and program, and more particularly, to a vanishing point estimation apparatus and program for estimating the position of a vanishing point in a captured image.

Conventionally, a method is known in which the intersection of each straight line pair is calculated from three straight lines acquired from an image, and the position of the vanishing point is calculated from the average position of the intersection (Patent Document 1).
JP 2000-123300 A

  However, in the technique described in Patent Document 1, when calculating the position of the vanishing point, it is necessary to always detect three straight lines such as white lines. The vanishing point position cannot be calculated, and there is a problem that the vanishing point position cannot be estimated stably.

  The present invention has been made to solve the above-described problem, and a vanishing point estimation apparatus and program capable of stably estimating the position of a vanishing point even when a plurality of straight lines are not detected. The purpose is to provide.

  In order to achieve the above object, a vanishing point estimation device according to a first aspect of the present invention is an imaging unit that captures a predetermined area at predetermined time intervals, and at least one straight line from each image captured by the imaging unit. A detecting means for detecting, a distribution representing positions of a plurality of hypothetical points assumed to exist on the image generated based on an image captured in the past by the imaging means, and a current imaging by the detecting means Updating means for updating the positions of the plurality of hypothesis points based on a distance between at least one straight line detected from the generated image and the positions of the hypothesis points, and generating the distribution; And estimation means for estimating the position of the vanishing point on the image based on the distribution generated by the means.

  According to a second aspect of the present invention, there is provided a program obtained by capturing a computer in the past by a detection unit that detects at least one straight line from each image captured by an imaging unit that captures a predetermined area at predetermined time intervals. A distribution representing the positions of a plurality of hypothetical points assumed to have vanishing points on the image generated based on the image, at least one straight line detected from the image picked up this time by the detecting means, and each Updating the positions of the plurality of hypothesis points based on the distance between the positions of the hypothesis points and generating the distribution; and on the image based on the distribution generated by the update means It is a program for functioning as an estimation means for estimating the position of the vanishing point.

  According to the first and second inventions, the imaging unit images a predetermined region at predetermined time intervals, and the detection unit detects at least one straight line from each image captured by the imaging unit.

  Then, the update means generates a distribution representing the positions of a plurality of hypothesis points on the image generated based on an image captured in the past by the imaging means, and at least one detected from the image captured this time by the detection means. Based on the distance between the straight line and the position of each hypothetical point, the positions of a plurality of hypothetical points are updated to generate a distribution.

  Then, the position of the vanishing point on the image is estimated by the estimation unit based on the distribution generated by the update unit.

  In this way, the positions of a plurality of hypothesis points are determined based on the distribution of the positions of hypothesis points generated based on images captured in the past and the distance between the detected at least one straight line and each hypothesis point. Update and generate hypothetical point position distribution, so even if multiple straight lines are not detected, hypothetical point position distribution can be generated and the vanishing point position can be estimated stably Can do.

  The estimation means according to the first invention can estimate the position of the vanishing point on the image based on the density of hypothesis points determined based on the distribution of the positions of the hypothesis points. Thereby, a position where the density of hypothesis points is high can be estimated as the position of the vanishing point.

  A vanishing point estimation device according to a third aspect of the present invention is an imaging unit that images a predetermined region at predetermined time intervals, a detection unit that detects at least one straight line from each image captured by the imaging unit, and the imaging A distribution representing a position of a plurality of hypothesis points assumed to exist on the image generated based on an image captured in the past by the means and a degree of existence of the vanishing point at each hypothesis point; and At least one of the positions of the plurality of hypothesis points and the degree of each hypothesis point based on the distance between at least one straight line detected from the image captured this time by the detection means and the position of each hypothesis point Updating means for generating the distribution, and estimating means for estimating the position of the vanishing point on the image based on the distribution generated by the updating means. To have.

  According to a fourth aspect of the present invention, there is provided a program for capturing a computer in the past by detection means for detecting at least one straight line from each image picked up by an image pickup means for picking up a predetermined area at predetermined time intervals. A distribution representing the positions of a plurality of hypothesis points on which the vanishing point is assumed to exist on the image generated based on the obtained image and the degree of vanishing points existing at the position of each hypothesis point, and this detection means Updating at least one of the positions of the plurality of hypothesis points and the degree of each hypothesis point based on a distance between at least one straight line detected from the captured image and the position of each hypothesis point; In order to function as an update unit that generates a distribution, and an estimation unit that estimates the position of the vanishing point on the image based on the distribution generated by the update unit Is a program.

  According to the third and fourth aspects of the invention, the imaging unit images a predetermined area at predetermined time intervals, and the detection unit detects at least one straight line from each image captured by the imaging unit.

  Then, a distribution indicating the position of a plurality of hypothesis points on the image generated based on the image captured in the past by the image capturing means and the degree of disappearance at each hypothesis point by the updating means, and detection Based on the distance between at least one straight line detected from the image captured this time by the means and the position of each hypothesis point, at least the degree of the existence of the positions of the plurality of hypothesis points and the vanishing points of each hypothesis point Update one to generate a distribution.

  Then, the position of the vanishing point on the image is estimated by the estimation unit based on the distribution generated by the update unit.

  As described above, the distribution representing the position of the hypothesis point generated based on the image captured in the past and the degree of the existence of the vanishing point, and the distance between the detected at least one straight line and each hypothesis point Even if multiple straight lines are not detected, the distribution of the position of the hypothesis points and the degree of existence of the vanishing points is updated to generate a distribution representing the position of the hypothesis points and the degree of existence of the vanishing points. The distribution representing the position of the hypothesis point and the degree to which the vanishing point exists can be generated, and the position of the vanishing point can be estimated stably.

  The updating means according to the third aspect of the invention can update the degree so that the hypothetical point having a shorter distance from the straight line detected by the detecting means becomes higher. As a result, the position of the hypothesis point that is close to the straight line is estimated as the position of the vanishing point.

  The estimating means according to the third invention can estimate the position of the vanishing point on the image based on at least one of the density of hypothesis points determined based on the distribution and the degree of each hypothesis point. Thereby, the position of the hypothesis point having a high density or a high degree of the hypothesis point can be estimated as the position of the vanishing point.

  The estimating means according to the third invention can estimate the position of the hypothesis point having the maximum degree as the position of the vanishing point.

  The estimating means according to the third invention may calculate the position of a point obtained by weighted averaging the positions of a plurality of hypothetical points using the degree as a weight, and estimate the calculated position of the point as the position of the vanishing point. it can.

  As described above, according to the vanishing point estimation device and the program of the present invention, even if a plurality of straight lines are not detected, the distribution of hypothetical point positions or the degree of the existence of hypothetical point positions and vanishing points Can be generated, and the vanishing point position can be estimated stably.

  Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. In the present embodiment, a case where the present invention is applied to a measurement apparatus that measures a distance to a measurement object will be described as an example.

  As shown in FIG. 1, a measurement apparatus 10 according to the first embodiment is mounted on a vehicle (not shown), and an imaging apparatus 12 that images a measurement target and generates an image, and the imaging apparatus 12 The computer 14 which stored the measurement program for implement | achieving the measurement process routine which measures the distance to a measurement object based on the image obtained from (1), and the display apparatus 16 which displays a measurement result are provided.

  The imaging device 12 continuously captures an area including a measurement object at a predetermined time interval and generates an image signal of the image. The imaging unit (not shown) includes an image signal generated by the imaging unit. A / D converter (not shown) for A / D converting the image signal, and an image memory (not shown) for temporarily storing the A / D converted image signal.

  The computer 14 includes a CPU, a ROM that stores a program of a measurement processing routine that will be described later, a RAM that stores data, and a bus that connects these. If the computer 14 is described in terms of functional blocks divided for each function realization means determined based on hardware and software, as shown in FIG. 1, a captured image that is a grayscale image at each time output from the imaging device 12. An image input unit 20 that inputs a feature point, a feature point extraction unit 22 that extracts a feature point from a captured image, a straight line detection unit 24 that detects a straight line from the captured image, and a plurality of vanishing points on the captured image. A distribution generation unit 26 that generates an existence degree distribution represented by the position of the hypothesis point and the degree of disappearance at the position of each hypothesis point, and a distribution storage unit 28 that stores the existence degree distribution generated in the past The vanishing point position estimating unit 30 that estimates the position of the vanishing point based on the presence degree distribution, and the measurement object detecting unit 32 that detects the measuring object from the captured image are estimated. Based on the position of the position and the detected measured object runs, and a distance measurement unit 34 which measures the distance from the imaging device 12 to the measurement object.

  The feature point extraction unit 22 extracts, for example, a luminance change point (edge point) from the captured image as a feature point. When extracting edge points, for example, feature points may be extracted using a differential filter such as a Sobel filter.

  The straight line detection unit 24 detects feature points arranged on a straight line and detects at least one straight line. As a method for detecting a straight line, a conventionally known method may be used. For example, a straight line may be detected from a captured image using Hough transform.

  As shown in FIG. 5, the presence degree distribution is represented by the positions of a plurality of hypothesis points where the vanishing points are assumed to exist and the weights as the degree of the vanishing points of the respective hypothesis points. The distribution generation unit 26 sets each hypothesis based on a hypothesis setting unit 38 that sets n hypothesis points on which the vanishing point is assumed to exist on the captured image and at least one straight line detected by the straight line detection unit 24. And an update unit 40 for updating the weight of the points.

  The hypothesis setting unit 38 sets n hypothesis points at a ratio proportional to the weight of each hypothesis point stored in the distribution storage unit 28. The update unit 40 updates the weights of n hypothetical points set on the captured image using the straight line detected by the straight line detection unit 24 as an observation vector. Based on the distance between the hypothesis point and the straight line, the update unit 40 updates the weight of each hypothesis point so that the shorter the distance, the greater the weight.

  The vanishing point position estimation unit 30 searches for the hypothesis point having the largest weight from the existence degree distribution represented by the n hypothesis points and weights generated by the distribution generation unit 26, and the position of the searched hypothesis point Is estimated as the position of the vanishing point.

  The measurement object detection unit 32 detects the measurement object from the captured image using, for example, a pattern matching method. The distance measuring unit 34 extends from the imaging device 12 to the measurement target based on the position of the measurement target detected by the measurement target detection unit 32 and the position of the vanishing point estimated by the vanishing point position estimation unit 30. Measure the distance. As a method for measuring the distance to the measurement object, a conventionally known method may be used, and in this embodiment, detailed description of the measurement method is omitted.

  Next, the operation of the measuring apparatus 10 according to the first embodiment will be described.

  First, when the imaging device 12 is directed to a region including a measurement target and the region including the measurement target is continuously imaged at a predetermined time interval by the imaging device 12, the computer 14 performs measurement processing illustrated in FIG. The routine is executed.

  In step 100, a variable t for identifying time is set to 0, which is an initial value. In step 102, the vanishing point position existence degree distribution is initialized, and the existence degree distribution is stored in the distribution storage unit 28.

Here, the existence degree distribution of the vanishing point position X _t (= (x _t , y _t )) at time t is represented by n hypothesis points s ⁱ _t and weights π ⁱ _t (i = 1, 1) for each hypothesis point. ..., n). Note that the set of hypothesis points is S _t = {s ⁰ _t ,..., S ⁿ _t }. The initialized presence degree distribution is a distribution represented by a predetermined hypothesis point position s ⁱ ₀ and a predetermined weight π ⁱ ₀ for each hypothesis point, and π ⁱ ₀ = 1 / n. That is, Σπ ⁱ ₀ = 1.

Then, in step 104, t is incremented, and in step 106, the existence degree distribution generated based on the image captured at the previous time (time t-1) is read from the distribution storage unit 28, and each hypothesis at time t-1 is read. The point s ⁱ _t-1 is transitioned according to the system model to generate a prediction hypothesis point s ⁱ _{t / t-1} at time t. Here, the system model is, for example, a model representing movement of the imaging device 12, and each hypothesis point is changed according to the movement of the imaging device 12.

  In the next step 108, a captured image captured at time t is acquired from the imaging device 12, and in step 110, a feature point that is an edge point is obtained from the captured image acquired in step 108 using a Sobel filter. Extract multiple.

In step 112, feature points arranged on a straight line are detected from the feature points extracted in step 110 using Hough transform, and at least one straight line is detected from the captured image. When the detected straight line group is an observation vector z _t , the observation vector z _t is expressed by the following equation (1).

Here, r ⁱ _t linearly in the captured image, m _t is the number of detected linear at time t.

In step 114, the weight π ⁱ _t (i = 1,..., N) at each of the prediction hypothesis points s ⁱ _{t / t−1} is updated using the observation vector z _t . In step 114, for each straight line r ^j _t (j = 1,..., M _t ), the prediction hypothesis point s ⁱ _{t / t−1} and the straight line r ^j _t are used to calculate w ^ij _t is calculated, and the weight π ⁱ _t for each prediction hypothesis point s ⁱ _{t / t−1} is updated by the following equation (2).

Here, w ^ij _t is the probability (likelihood) of obtaining the observation vector r ^j _t when the vanishing point position X _t is the position of the prediction hypothesis point s ⁱ _t . For example, as shown in FIG. 3, when the distance between the predicted hypothesis point s ⁱ _t and the observation vector r ^j _t representing the straight line is d, w ^ij _t is obtained using a function as shown in FIG. As shown in FIG. 4, a prediction hypothesis point that is closer to the detected straight line is updated so that the corresponding weight is a larger weight, while a prediction hypothesis point that is farther from the detected straight line is updated. The corresponding weight is updated so as to be a small weight. Also, π ⁱ _t is normalized so as to satisfy the following expression (3).

  In the next step 116, the presence degree distribution represented by the position of the prediction hypothesis point generated in step 106 and the weight for each prediction hypothesis point updated in step 114 is stored in the distribution storage unit 28. 5, the hypothesis point having the maximum weight is searched from the presence distribution, and the position of the searched hypothesis point on the captured image is estimated as the position of the vanishing point.

  In step 120, the measurement object is detected from the captured image acquired in step 108. In step 122, the position of the vanishing point on the captured image estimated in step 118 and detected in step 120. Based on the position of the measured object on the captured image, the distance from the imaging device 12 to the measurement object is measured, and in step 124, the distance to the measurement object measured in step 122 is displayed. It is displayed on the device 16.

  In the next step 126, it is determined whether or not the measurement process is to be ended. If the operator has instructed to end the measurement process, the measurement process routine is ended. On the other hand, there has been no instruction to end the measurement process. In this case, in step 128, the presence degree distribution stored in the distribution storage unit 28 is updated, the process returns to step 104, and the processing of steps 104 to 124 is executed for the captured image captured at time t + 1. To do.

Updating of step 128, at a rate proportional to the weight .pi.i t for each of the n predicted hypotheses point s i t / t-1, then extracted with n hypotheses point from the captured image, the n hypotheses The position of the point is updated to obtain a set of hypothetical points S t = {s 1 t ,..., S n t }. The weight of each hypothesis point is again set to π i t = 1 / n.

  As described above, according to the measurement apparatus according to the first embodiment, the presence degree distribution represented by the position and weight of the hypothesis point generated based on the previously captured image, and at least one detected Based on the distance between the line and each hypothesis point, the weight distribution for each hypothesis point is updated to generate the presence distribution, so even if multiple straight lines are not detected, the presence distribution is generated. And the position of the vanishing point can be stably estimated using the existence degree distribution. In addition, the distance to the measurement object can be stably measured using the estimated vanishing point position.

  Even in the prior art, it is possible to obtain the current estimated value using the past estimated value, but the estimation accuracy is low because the information acquired from the current captured image is not effectively used. On the other hand, in the present invention, since the presence degree distribution is generated using at least one straight line detected from the current captured image, the vanishing point position is determined by effectively using the information acquired from the current captured image. It can be estimated with high accuracy.

  In addition, the existence distribution used to estimate the position of the vanishing point in the captured image is expressed by the position of multiple hypothesis points and the weight for each hypothesis point, and the weight of each hypothesis point based on the information of the detected straight line Is updated to generate the existence degree distribution, and the position of the vanishing point is estimated from the existence degree distribution of the vanishing point. This makes it possible to generate a stable presence distribution by evaluating the weights of hypothesis points based on the positional relationship between each hypothesis point and a straight line, without directly calculating the intersection coordinates of the straight lines that are candidates for the position of the vanishing point. Therefore, it is possible to stabilize the estimation of the vanishing point position.

  In addition, even when a straight line pair whose vanishing point is an intersection is not detected, the weight of the hypothesis point can be evaluated, and the estimation of the vanishing point position is stabilized.

In the above embodiment, the case where the presence degree distribution is represented by the position of the hypothesis point and the weight for each hypothesis point has been described as an example. However, the presence degree distribution is represented by the position of the hypothesis point and each hypothesis point. You may make it represent with the presence probability in which a vanishing point exists. In this case, when the observation vector history is Z _t = {z ₁ ,..., Z _t }, the existence probability distribution p (X _t | Z _t ) is changed to n discrete states. What is necessary is just to calculate using hypothesis point s ⁱ _t and weight π ⁱ _t for each hypothesis point. Further, when estimating the position of the vanishing point, the position of the hypothesis point having the maximum existence probability may be searched, and the position of the searched hypothesis point may be estimated as the position of the vanishing point.

  In addition, the case where the weight of each hypothesis point is calculated using the distance between the detected line and the position of each hypothesis point has been described as an example, but the number of feature points existing on the detected line is further considered. Then, the weight of each hypothesis point may be calculated. For example, when a plurality of straight lines are detected, a large weight is calculated for a hypothetical point close to a straight line with a large number of feature points existing on the straight line, while close to a straight line with a small number of feature points existing on the straight line. A small weight is calculated for the hypothesis point.

  Next, a second embodiment will be described. In addition, since the structure of the measuring device which concerns on 2nd Embodiment is the structure similar to 1st Embodiment, it attaches | subjects the same code | symbol and abbreviate | omits description.

  The second embodiment is different from the first embodiment in that the position of the vanishing point is estimated based on the density of hypothesis points determined from the presence degree distribution and the weight of each hypothesis point.

  In the process of estimating the position of the vanishing point in the measurement processing routine according to the second embodiment, the density of hypothesis points determined from the presence degree distribution is calculated for the position of each hypothesis point, and the value obtained by multiplying the weight by the density The position of the hypothesis point that maximizes is estimated as the position of the vanishing point.

  For example, in step 128 of the measurement processing routine according to the first embodiment described above, as a result of extracting n hypothesis points at a proportion proportional to the weight for each of the predicted hypothesis points, a plurality of hypothesis points are the same. When extracted at a position, the density of the hypothesis point at that position becomes high, and if the weight of the hypothesis point at the position where the hypothesis point density is high is large, the position of this hypothesis point becomes the position of the vanishing point. The possibility of being estimated increases.

  In this way, based on the density of hypothesis points determined from the existence distribution and the weight of each hypothesis point, the position of the hypothesis point having a high hypothesis point density and a large weight is estimated. Thus, the position of the vanishing point can be estimated with high accuracy.

  Next, a third embodiment will be described. In addition, since the structure of the measuring device which concerns on 3rd Embodiment is the structure similar to 1st Embodiment, it attaches | subjects the same code | symbol and abbreviate | omits description.

  The third embodiment is different from the first embodiment in that a position calculated by a weighted average of hypothetical point positions is estimated as a vanishing point position.

  In the process of estimating the position of the vanishing point in the measurement processing routine according to the third embodiment, as shown in FIG. 6, based on the position of each hypothesis point in the presence degree distribution and the weight for each hypothesis point, A weighted average of hypothetical point positions is calculated, and the calculated position is estimated as the position of the vanishing point.

  Thus, by estimating the position calculated by the weighted average of the positions of the hypothesis points as the position of the vanishing point, the position of the vanishing point can be estimated with high accuracy.

  Next, a fourth embodiment will be described. In addition, since the structure of the measuring device which concerns on 4th Embodiment is the structure similar to 1st Embodiment, it attaches | subjects the same code | symbol and abbreviate | omits description.

  In the fourth embodiment, the point where the presence degree distribution is represented by the distribution of the vanishing point position and the point where the position of the vanishing point is estimated based on the density of hypothesis points determined from the presence degree distribution are: This is different from the first embodiment.

  The distribution generation unit 26 of the measurement apparatus 10 according to the fourth embodiment generates an existence degree distribution represented by a distribution of n hypothetical point positions. A weight is set for each hypothesis point, and the density of hypothesis points determined from the presence distribution is obtained for each position on the captured image.

  The vanishing point position estimation unit 30 estimates a position having the highest density of hypothesis points determined from the presence degree distribution as the position of the vanishing point.

  Next, a measurement processing routine according to the fourth embodiment will be described with reference to FIG. In addition, about the process similar to 1st Embodiment, the same code | symbol is attached | subjected and detailed description is abbreviate | omitted.

  First, in step 100, a variable t for identifying time is set to 0, and in step 102, the vanishing point position existence degree distribution is initialized and stored in the distribution storage unit 28.

Here, the existence degree distribution of the vanishing point position X _t (= (x _t , y _t )) at time t is expressed by the distribution of the positions of n hypothetical points s ⁱ _t . The initialized presence degree distribution is a distribution of predetermined hypothesis point positions s ⁱ ₀ , and a predetermined weight π ⁱ ₀ is set for each hypothesis point. Note that π ⁱ ₀ = 1 / n, that is, Σπ ⁱ ₀ = 1.

Then, in step 104, t is incremented, and in step 106, the existence degree distribution generated based on the image captured at the previous time (time t-1) is read from the distribution storage unit 28, and each hypothesis at time t-1 is read. The point s ⁱ _t-1 is transitioned according to the system model to generate a prediction hypothesis point s ⁱ _{t / t-1} at time t. In the next step 108, a captured image captured at time t is acquired from the imaging device 12, and in step 110, a plurality of feature points that are edge points are extracted from the captured image acquired in step 108.

In step 112, at least one straight line is detected from the captured image, and in step 114, each predicted hypothesis point s ⁱ _{t / t−1} is set using the observation vector z _t representing the detected straight line. Updated weights π ⁱ _t (i = 1,..., N).

In the next step 400, n hypothesis points are extracted from the captured image at a ratio proportional to the weight π ⁱ _t for each of the n prediction hypothesis points s ⁱ _{t / t−1} , and n hypotheses are obtained. The position of the point is updated to obtain a set of hypothetical points S _t = {s ¹ _t ,..., S ⁿ _t }. Further, the weight of each hypothesis point is set again to π ⁱ _t = 1 / n.

  In step 402, the existence degree distribution represented by the distribution of the hypothesis point positions updated in step 400 is stored in the distribution storage unit 28. In step 404, the density of hypothesis points determined from the existence degree distribution is imaged. While calculating for each position on the image, a position where the density of hypothesis points is maximized is searched, and the searched position is estimated as the position of the vanishing point on the captured image.

  In step 120, the measurement object is detected from the captured image acquired in step 108. In step 122, the position of the vanishing point on the captured image estimated in step 404 and the position detected in step 120 are detected. Based on the position of the measured object on the captured image, the distance from the imaging device 12 to the measurement object is measured, and in step 124, the distance to the measurement object measured in step 122 is displayed. It is displayed on the device 16.

  In the next step 126, it is determined whether or not the measurement process is to be ended. If the operator has instructed to end the measurement process, the measurement process routine is ended. On the other hand, there has been no instruction to end the measurement process. If so, the process returns to step 104 above.

  As described above, according to the measurement apparatus according to the fourth embodiment, the presence degree distribution represented by the distribution of the positions of the hypothesis points generated based on the previously captured image, and at least one detected Based on the distance between the straight line and each hypothetical point, the position of multiple hypothetical points is updated to generate a presence distribution, so even if multiple straight lines are not detected, the degree of vanishing point presence A distribution can be generated, and the position of the vanishing point can be stably estimated using the presence degree distribution.

It is a block diagram which shows the measuring device which concerns on the 1st Embodiment of this invention. It is a flowchart which shows the content of the measurement process routine in the measuring device which concerns on the 1st Embodiment of this invention. It is an image figure which shows the distance between the detected straight line and a hypothesis point. It is a graph which shows the relationship between the distance between a hypothesis point and a straight line, and the weight with respect to a hypothesis point. It is an image figure which shows presence degree distribution. It is an image figure which shows a mode that the weighted average of the position of a hypothesis point is calculated from presence degree distribution. It is a flowchart which shows the content of the measurement process routine in the measuring device which concerns on the 4th Embodiment of this invention.

Explanation of symbols

DESCRIPTION OF SYMBOLS 10 Measuring apparatus 12 Imaging apparatus 14 Computer 22 Feature point extraction part 24 Straight line detection part 26 Distribution generation part 28 Distribution memory | storage part 30 Vanishing point position estimation part 32 Measurement target object detection part 34 Distance measurement part

Claims (9)

Imaging means for imaging a predetermined area at predetermined time intervals;
Detection means for detecting at least one straight line from each image picked up by the image pickup means;
A distribution representing the positions of a plurality of hypothesis points on the image generated based on an image captured in the past by the imaging means, and detected from the image captured this time by the detection means. Updating means for updating the positions of the plurality of hypothesis points based on a distance between at least one straight line and the positions of the hypothesis points;
Estimating means for estimating the position of the vanishing point on the image based on the distribution generated by the updating means;
Vanishing point estimation apparatus including:   The vanishing point estimation device according to claim 1, wherein the estimation unit estimates a position of a vanishing point on the image based on a density of hypothesis points determined based on the distribution. Imaging means for imaging a predetermined area at predetermined time intervals;
Detection means for detecting at least one straight line from each image picked up by the image pickup means;
A distribution representing the positions of a plurality of hypothesis points on which it is assumed that there are vanishing points on the image generated based on images captured in the past by the imaging means, and the degree of vanishing points at the positions of the respective hypothesis points And the position of the plurality of hypothesis points and the degree of each hypothesis point based on the distance between at least one straight line detected from the image captured this time by the detection means and the position of each hypothesis point. Updating means for updating at least one to generate the distribution;
Estimating means for estimating the position of the vanishing point on the image based on the distribution generated by the updating means;
Vanishing point estimation apparatus including:   The vanishing point estimation apparatus according to claim 3, wherein the update unit updates the degree so that a hypothesis point closer to the straight line detected by the detection unit becomes higher.   The vanishing point estimation according to claim 3 or 4, wherein the estimating means estimates the position of the vanishing point on the image based on at least one of a density of hypothesis points determined based on the distribution and the degree of each hypothesis point. apparatus.   The vanishing point estimation device according to any one of claims 3 to 5, wherein the estimating means estimates a position of a hypothesis point having the maximum degree as a vanishing point position.   The estimation means calculates a position of a point obtained by weighted averaging the positions of the plurality of hypothesis points using the degree as a weight, and estimates the calculated position of the point as a position of a vanishing point. Item 6. The vanishing point estimation device according to any one of items 5 to 6. Computer
Detecting means for detecting at least one straight line from each image picked up by an image pickup means for picking up a predetermined region at predetermined time intervals;
A distribution representing the positions of a plurality of hypothetical points on which it is assumed that there are vanishing points on the image generated based on an image captured in the past by the imaging unit, and detected from the image currently captured by the detection unit. Updating the plurality of hypothesis points based on the distance between at least one straight line and the position of each hypothesis point, and generating the distribution, and the update unit generated by the update unit A program for functioning as an estimation means for estimating the position of a vanishing point on the image based on the distribution. Computer
Detecting means for detecting at least one straight line from each image picked up by an image pickup means for picking up a predetermined region at predetermined time intervals;
A distribution representing the positions of a plurality of hypothesis points that are assumed to have vanishing points on the image generated based on images captured in the past by the imaging means and the degree of vanishing points at the positions of the respective hypothesis points And the position of the plurality of hypothesis points and the degree of each hypothesis point based on the distance between at least one straight line detected from the image captured this time by the detection means and the position of each hypothesis point. An update means for updating at least one to generate the distribution, and a program for functioning as an estimation means for estimating a position of a vanishing point on the image based on the distribution generated by the update means.

Images