How to find intersection of straight lines with Hough Peaks?

.everyoneloves__top-leaderboard:empty,.everyoneloves__mid-leaderboard:empty,.everyoneloves__bot-mid-leaderboard:empty{ height:90px;width:728px;box-sizing:border-box;
}

I am developing a method to localise license plate and ultimately apply perspective transform to straighten the plate horizontally.

Currently, my image processing pipeline is able to roughly locate the plate but I'd like to narrow it further. I have applied edge detection, morphological operations and connected component analysis to get to this stage.

My result for license plate detection currently is this

enter image description here

I am trying to implement HoughLinesP to find straight lines of the fonts and then find the intersection of the lines with HoughPeaks, like this

enter image description here

My current attempt at finding houghlines is shown at the code below. However, can anyone guide me on how to find hough peaks? I did some research online and couldn't find any reliable reference.

def findHoughLines(self, image):

    from src.Line import Line



    # Add 1 channel to binary image

    image = image[..., np.newaxis]

    vis = cv2.cvtColor(image, cv2.COLOR_GRAY2BGR)



    canny = cv2.Canny(image, 100, 200)



    cv2.imshow("canny", canny)



    lines = cv2.HoughLinesP(canny, rho=0.01, theta=np.pi/360, threshold=10, minLineLength=100, maxLineGap=30)



    print(len(lines))

    linesHorizontal, linesVertical = , 



    if lines is None:

        print("[Info] No Hough lines have been detected")

        return image



    for line in lines:

        x1, y1, x2, y2 = line[0]

        if abs(x1 - x2) > abs(y1 - y2):

            linesHorizontal.append(Line(line[0]))

        else:

            linesVertical.append(Line(line[0]))



    # sort lines

    linesHorizontal = sorted(linesHorizontal, key=lambda l: l.d_y, reverse=True)[:10]

    linesVertical = sorted(linesVertical, key=lambda l: l.d_x, reverse=True)[:10]



    for line in linesHorizontal:

        x1, y1, x2, y2 = line.points

        cv2.line(vis, (x1, y1), (x2, y2), (0, 0, 255), 1)



    for line in linesVertical:

        x1, y1, x2, y2 = line.points

        cv2.line(vis, (x1, y1), (x2, y2), (0, 255, 0), 1)



    cv2.imshow("HoughLines", vis)

Result of HoughLinesP Detection

enter image description here

edited Nov 22 '18 at 9:53

asked Nov 22 '18 at 9:27

Moore Tech

456

Hough peaks is just the peaks in the hough matrix, or in other words the ones that passes certain amount of votes in the accumulator... In HoughLinesP this is already done, and it returns the lines which pass the threshold of votes given in the function (in your code you give 5)

– api55
Nov 22 '18 at 9:58

Thanks for your clarification. Do you think there is a better approach to my problem?

– Moore Tech
Nov 22 '18 at 10:02

I haven't done much in the license plate recognition area, but I think that you can try to find contours, and get the bottom coordinate of each contour, and try to create a line that fits (with certain tolerance) to the bottom points of each contour (maybe with RANSAC), do the same with top of the contours, the letters/numbers have a fixed height and you can use this baseline to know which contours are letters. Then you put all the points together and use minAreaRect (or something like that) which gives you even the angle that you need to rotate the image to make it straight

– api55
Nov 22 '18 at 10:14

Thanks! That’s very insightful! Will try it out

– Moore Tech
Nov 22 '18 at 11:13

add a comment |

I am developing a method to localise license plate and ultimately apply perspective transform to straighten the plate horizontally.

My result for license plate detection currently is this

enter image description here

I am trying to implement HoughLinesP to find straight lines of the fonts and then find the intersection of the lines with HoughPeaks, like this

enter image description here

My current attempt at finding houghlines is shown at the code below. However, can anyone guide me on how to find hough peaks? I did some research online and couldn't find any reliable reference.

def findHoughLines(self, image):

    from src.Line import Line



    # Add 1 channel to binary image

    image = image[..., np.newaxis]

    vis = cv2.cvtColor(image, cv2.COLOR_GRAY2BGR)



    canny = cv2.Canny(image, 100, 200)



    cv2.imshow("canny", canny)



    lines = cv2.HoughLinesP(canny, rho=0.01, theta=np.pi/360, threshold=10, minLineLength=100, maxLineGap=30)



    print(len(lines))

    linesHorizontal, linesVertical = , 



    if lines is None:

        print("[Info] No Hough lines have been detected")

        return image



    for line in lines:

        x1, y1, x2, y2 = line[0]

        if abs(x1 - x2) > abs(y1 - y2):

            linesHorizontal.append(Line(line[0]))

        else:

            linesVertical.append(Line(line[0]))



    # sort lines

    linesHorizontal = sorted(linesHorizontal, key=lambda l: l.d_y, reverse=True)[:10]

    linesVertical = sorted(linesVertical, key=lambda l: l.d_x, reverse=True)[:10]



    for line in linesHorizontal:

        x1, y1, x2, y2 = line.points

        cv2.line(vis, (x1, y1), (x2, y2), (0, 0, 255), 1)



    for line in linesVertical:

        x1, y1, x2, y2 = line.points

        cv2.line(vis, (x1, y1), (x2, y2), (0, 255, 0), 1)



    cv2.imshow("HoughLines", vis)

Result of HoughLinesP Detection

enter image description here

edited Nov 22 '18 at 9:53

asked Nov 22 '18 at 9:27

Moore Tech

456

Hough peaks is just the peaks in the hough matrix, or in other words the ones that passes certain amount of votes in the accumulator... In HoughLinesP this is already done, and it returns the lines which pass the threshold of votes given in the function (in your code you give 5)

– api55
Nov 22 '18 at 9:58

Thanks for your clarification. Do you think there is a better approach to my problem?

– Moore Tech
Nov 22 '18 at 10:02

I haven't done much in the license plate recognition area, but I think that you can try to find contours, and get the bottom coordinate of each contour, and try to create a line that fits (with certain tolerance) to the bottom points of each contour (maybe with RANSAC), do the same with top of the contours, the letters/numbers have a fixed height and you can use this baseline to know which contours are letters. Then you put all the points together and use minAreaRect (or something like that) which gives you even the angle that you need to rotate the image to make it straight

– api55
Nov 22 '18 at 10:14

Thanks! That’s very insightful! Will try it out

– Moore Tech
Nov 22 '18 at 11:13

add a comment |

I am developing a method to localise license plate and ultimately apply perspective transform to straighten the plate horizontally.

My result for license plate detection currently is this

enter image description here

I am trying to implement HoughLinesP to find straight lines of the fonts and then find the intersection of the lines with HoughPeaks, like this

enter image description here

My current attempt at finding houghlines is shown at the code below. However, can anyone guide me on how to find hough peaks? I did some research online and couldn't find any reliable reference.

def findHoughLines(self, image):

    from src.Line import Line



    # Add 1 channel to binary image

    image = image[..., np.newaxis]

    vis = cv2.cvtColor(image, cv2.COLOR_GRAY2BGR)



    canny = cv2.Canny(image, 100, 200)



    cv2.imshow("canny", canny)



    lines = cv2.HoughLinesP(canny, rho=0.01, theta=np.pi/360, threshold=10, minLineLength=100, maxLineGap=30)



    print(len(lines))

    linesHorizontal, linesVertical = , 



    if lines is None:

        print("[Info] No Hough lines have been detected")

        return image



    for line in lines:

        x1, y1, x2, y2 = line[0]

        if abs(x1 - x2) > abs(y1 - y2):

            linesHorizontal.append(Line(line[0]))

        else:

            linesVertical.append(Line(line[0]))



    # sort lines

    linesHorizontal = sorted(linesHorizontal, key=lambda l: l.d_y, reverse=True)[:10]

    linesVertical = sorted(linesVertical, key=lambda l: l.d_x, reverse=True)[:10]



    for line in linesHorizontal:

        x1, y1, x2, y2 = line.points

        cv2.line(vis, (x1, y1), (x2, y2), (0, 0, 255), 1)



    for line in linesVertical:

        x1, y1, x2, y2 = line.points

        cv2.line(vis, (x1, y1), (x2, y2), (0, 255, 0), 1)



    cv2.imshow("HoughLines", vis)

Result of HoughLinesP Detection

enter image description here

edited Nov 22 '18 at 9:53

asked Nov 22 '18 at 9:27

Moore Tech

456

I am developing a method to localise license plate and ultimately apply perspective transform to straighten the plate horizontally.

My result for license plate detection currently is this

enter image description here

I am trying to implement HoughLinesP to find straight lines of the fonts and then find the intersection of the lines with HoughPeaks, like this

enter image description here

My current attempt at finding houghlines is shown at the code below. However, can anyone guide me on how to find hough peaks? I did some research online and couldn't find any reliable reference.

def findHoughLines(self, image):

    from src.Line import Line



    # Add 1 channel to binary image

    image = image[..., np.newaxis]

    vis = cv2.cvtColor(image, cv2.COLOR_GRAY2BGR)



    canny = cv2.Canny(image, 100, 200)



    cv2.imshow("canny", canny)



    lines = cv2.HoughLinesP(canny, rho=0.01, theta=np.pi/360, threshold=10, minLineLength=100, maxLineGap=30)



    print(len(lines))

    linesHorizontal, linesVertical = , 



    if lines is None:

        print("[Info] No Hough lines have been detected")

        return image



    for line in lines:

        x1, y1, x2, y2 = line[0]

        if abs(x1 - x2) > abs(y1 - y2):

            linesHorizontal.append(Line(line[0]))

        else:

            linesVertical.append(Line(line[0]))



    # sort lines

    linesHorizontal = sorted(linesHorizontal, key=lambda l: l.d_y, reverse=True)[:10]

    linesVertical = sorted(linesVertical, key=lambda l: l.d_x, reverse=True)[:10]



    for line in linesHorizontal:

        x1, y1, x2, y2 = line.points

        cv2.line(vis, (x1, y1), (x2, y2), (0, 0, 255), 1)



    for line in linesVertical:

        x1, y1, x2, y2 = line.points

        cv2.line(vis, (x1, y1), (x2, y2), (0, 255, 0), 1)



    cv2.imshow("HoughLines", vis)

Result of HoughLinesP Detection

enter image description here

python opencv image-processing houghlinesp

edited Nov 22 '18 at 9:53

asked Nov 22 '18 at 9:27

Moore Tech

456

edited Nov 22 '18 at 9:53

asked Nov 22 '18 at 9:27

Moore Tech

456

edited Nov 22 '18 at 9:53

asked Nov 22 '18 at 9:27

Moore Tech

456

asked Nov 22 '18 at 9:27

Moore Tech

456

asked Nov 22 '18 at 9:27

Moore Tech

456

Hough peaks is just the peaks in the hough matrix, or in other words the ones that passes certain amount of votes in the accumulator... In HoughLinesP this is already done, and it returns the lines which pass the threshold of votes given in the function (in your code you give 5)

– api55
Nov 22 '18 at 9:58

Thanks for your clarification. Do you think there is a better approach to my problem?

– Moore Tech
Nov 22 '18 at 10:02

I haven't done much in the license plate recognition area, but I think that you can try to find contours, and get the bottom coordinate of each contour, and try to create a line that fits (with certain tolerance) to the bottom points of each contour (maybe with RANSAC), do the same with top of the contours, the letters/numbers have a fixed height and you can use this baseline to know which contours are letters. Then you put all the points together and use minAreaRect (or something like that) which gives you even the angle that you need to rotate the image to make it straight

– api55
Nov 22 '18 at 10:14

Thanks! That’s very insightful! Will try it out

– Moore Tech
Nov 22 '18 at 11:13

add a comment |

Hough peaks is just the peaks in the hough matrix, or in other words the ones that passes certain amount of votes in the accumulator... In HoughLinesP this is already done, and it returns the lines which pass the threshold of votes given in the function (in your code you give 5)

– api55
Nov 22 '18 at 9:58

Thanks for your clarification. Do you think there is a better approach to my problem?

– Moore Tech
Nov 22 '18 at 10:02

I haven't done much in the license plate recognition area, but I think that you can try to find contours, and get the bottom coordinate of each contour, and try to create a line that fits (with certain tolerance) to the bottom points of each contour (maybe with RANSAC), do the same with top of the contours, the letters/numbers have a fixed height and you can use this baseline to know which contours are letters. Then you put all the points together and use minAreaRect (or something like that) which gives you even the angle that you need to rotate the image to make it straight

– api55
Nov 22 '18 at 10:14

Thanks! That’s very insightful! Will try it out

– Moore Tech
Nov 22 '18 at 11:13

Hough peaks is just the peaks in the hough matrix, or in other words the ones that passes certain amount of votes in the accumulator... In HoughLinesP this is already done, and it returns the lines which pass the threshold of votes given in the function (in your code you give 5)

– api55
Nov 22 '18 at 9:58

Thanks for your clarification. Do you think there is a better approach to my problem?

– Moore Tech
Nov 22 '18 at 10:02

I haven't done much in the license plate recognition area, but I think that you can try to find contours, and get the bottom coordinate of each contour, and try to create a line that fits (with certain tolerance) to the bottom points of each contour (maybe with RANSAC), do the same with top of the contours, the letters/numbers have a fixed height and you can use this baseline to know which contours are letters. Then you put all the points together and use minAreaRect (or something like that) which gives you even the angle that you need to rotate the image to make it straight

– api55
Nov 22 '18 at 10:14

Thanks! That’s very insightful! Will try it out

– Moore Tech
Nov 22 '18 at 11:13

add a comment |

0

active

oldest

votes

Your Answer

StackExchange.ifUsing("editor", function () {
StackExchange.using("externalEditor", function () {
StackExchange.using("snippets", function () {
StackExchange.snippets.init();
});
});
}, "code-snippets");

StackExchange.ready(function() {
var channelOptions = {
tags: "".split(" "),
id: "1"
};
initTagRenderer("".split(" "), "".split(" "), channelOptions);

StackExchange.using("externalEditor", function() {
// Have to fire editor after snippets, if snippets enabled
if (StackExchange.settings.snippets.snippetsEnabled) {
StackExchange.using("snippets", function() {
createEditor();
});
}
else {
createEditor();
}
});

function createEditor() {
StackExchange.prepareEditor({
heartbeatType: 'answer',
autoActivateHeartbeat: false,
convertImagesToLinks: true,
noModals: true,
showLowRepImageUploadWarning: true,
reputationToPostImages: 10,
bindNavPrevention: true,
postfix: "",
imageUploader: {
brandingHtml: "Powered by u003ca class="icon-imgur-white" href="https://imgur.com/"u003eu003c/au003e",
contentPolicyHtml: "User contributions licensed under u003ca href="https://creativecommons.org/licenses/by-sa/3.0/"u003ecc by-sa 3.0 with attribution requiredu003c/au003e u003ca href="https://stackoverflow.com/legal/content-policy"u003e(content policy)u003c/au003e",
allowUrls: true
},
onDemand: true,
discardSelector: ".discard-answer"
,immediatelyShowMarkdownHelp:true
});

}
});

draft saved

draft discarded

Sign up or log in

StackExchange.ready(function () {
StackExchange.helpers.onClickDraftSave('#login-link');
});

Post as a guest

Name

Required, but never shown

StackExchange.ready(
function () {
StackExchange.openid.initPostLogin('.new-post-login', 'https%3a%2f%2fstackoverflow.com%2fquestions%2f53427645%2fhow-to-find-intersection-of-straight-lines-with-hough-peaks%23new-answer', 'question_page');
}
);

Post as a guest

Name

Required, but never shown

0

active

oldest

votes

0

active

oldest

votes

draft saved

draft discarded

Thanks for contributing an answer to Stack Overflow!

Please be sure to answer the question. Provide details and share your research!

But avoid …

Asking for help, clarification, or responding to other answers.

Making statements based on opinion; back them up with references or personal experience.

To learn more, see our tips on writing great answers.

draft saved

draft discarded

Sign up or log in

StackExchange.ready(function () {
StackExchange.helpers.onClickDraftSave('#login-link');
});

Post as a guest

Name

Required, but never shown

Post as a guest

Name

Required, but never shown

Sign up or log in

StackExchange.ready(function () {
StackExchange.helpers.onClickDraftSave('#login-link');
});

Post as a guest

Name

Required, but never shown

Sign up or log in

StackExchange.ready(function () {
StackExchange.helpers.onClickDraftSave('#login-link');
});

Post as a guest

Name

Required, but never shown

Sign up or log in

StackExchange.ready(function () {
StackExchange.helpers.onClickDraftSave('#login-link');
});

Post as a guest

Name

Required, but never shown

Name

Required, but never shown

Name

Required, but never shown

This page is only for reference, If you need detailed information, please check here

搜尋此網誌

Agfdhyk