Dear colleagues,
I am new in the area of computer vision and, in particular, extraction of useful features from an object. I have tried to segment an image, as shown in the attachment. The image is a chip, and I need to put a border around it. The challenge is that the only features available to me are lead-pin objects; all other regions contain non-feature objects. I tried the matlab command, bwboundaries(.), but it is unable to address my issue. Anyone with idea please?
You can try imcontour, bwperim.
And let me know whether they work for you or not.
Thanks for the response, Neha. I have tried the two functions you have suggested. However, the functions trace the boundaries of the features. I also have tried to change the input parameters of the functions, but are still inadequately engulfing the chip borders. Please find below the attachment with the results I am getting. The left figure is for the imcontour() function.
I have made some progress to insert a bounding box, using a matlab command "rectangle('Position', rect, 'EdgeColor', 'blue');". This function works well for unrotated chips. If the chip is rotated, however, I see that the function doesn't work well (unless there are some options I am missing out). The attachment shows the results I get when the chip is rotated.
So you can try morphological function imdilate, to dilate image first an then apply boundary function.
And try below function to get result by imboundaries.
Thank you for the nice suggestion, Neha. I have tried the procedures, but the final image doesn't have the bounding-box; only high-valued features are highlighted. When I was surfing, I got a particular code that seems to address the problem. Running the code draws a border around the chip. I am still working on the code to see how I may improve it further -- I need to locate the position of the center of the chip; meaning (x, y, theta). Let me share the code and the results I get so far.
- Badges
- Science topic