how to connect the detection object

Question

yasmin ismail 2023-7-11

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/1994483-how-to-connect-the-detection-object

回答： Image Analyst 2023-7-12

I am biggener in using matlab , i used different filter and thresholding methods but stil discountiuity in object detection as shown in attach imeges

Thus anyone have suggestion to fix this issues, please

2 个评论
显示无隐藏无

Rahul 2023-7-11

Hi Yasmin,

Could you give share more details regarding what exactly is your aim.

According to the photo shared I believe that edge detection methods can detect the inconsistant line.

However, if you want to detect some other objects, exploring the Computer Vision toolbox from Mathworks might help. Computer Vision Toolbox - MATLAB (mathworks.com)

Thanks.

yasmin ismail 2023-7-11

sav.jpg

@Rahul the attached image has marker circle indicate where the discontiuity, the crack should be contioues as original image but I tried different filter and thresholding but couldnt make the detection same as original , can you help me to fix this issue?

请先登录，再进行评论。

请先登录，再回答此问题。

Answer 1

Kush 2023-7-11

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/1994483-how-to-connect-the-detection-object#answer_1271488

在 MATLAB Online 中打开

You can try multple filters such as sobel, prewitt etc given in edge function of matlab, I used color thresholding and box filter to get the following result.The colour thresholding helps capture only the dark pixels since the cracks would be darker and the box filter helps connect broken edges by averaging surrounding value to a center pixel during convolution.

image = imread('inputImage.jpeg'); %Insert your input image filename here
hsvImage = rgb2hsv(image);
valueChannel = hsvImage(:, :, 3);
threshold = 0.3; % Adjust this value as needed
binaryMask = valueChannel < threshold;
convolvedImage = imboxfilt(double(binaryMask), [3,3]); %Choose a window size for blurring
finalImage = ~convolvedImage %Invert the output if necessary
imshow(finalImage);
title('Dark Color Image');