Can computer see?Surely not.Thiswould be the answer if we were living a decade back. In today’s worldvarious algorithms are available by which we can make a computer interpretthings about an image, 4 way technologies.We can make a computer not only detect the various object’s in the image but we can make it recognize those object. There is the slight difference between object detection and object recognition, in object detection the computer only detects that there are various object present in the image, but object recognition takes this to the next level.In this the computer not only detects various objects but also detects the various features of each object in image which allow it to distinguish between different object. There are various algorithms available to do both object detection and object recognisation.
How to do it :-
So you might be wondering it is a very complicated task?right ?
Lets see it all in action.
If you are from the coding background then you might have learnt “array” at some point in your
college or school. Well,computerinterpretimages as the array,(array of array) to be more specific.
The figure on the left is the image given to the computer and the figure on the right shows how the
computer interprets that image. Clearly one can see that computer is interpreting an image as an
array of array. Dimension of the image an array and the image are same (for obvious reason) since the
array is how computer looks at the image hence the size of both image and array is the same and each
element of the array denotes the intensity of that pixel.
There are various libraries to do the task mentioned above . One of the most developed and famous one
isOpenCV(Open Computer Vision) most commonly used by us at 4Way Technologies. This library contain lots of predefined functions to do image
processing and this library is available on most of the famous programming platform in the world
including languages like c++ , Python ,Java(JavaCV).
The part of computer science which deals with the problems related as discussed above is called Image Processing. We can make almost a billion images from a given image,but only some of the images are correct image processing helps us in detecting the correct image.
Image processing deals in continuously developing techniques to make “computer vision” better and better, using techniques of AI or by developing new algorithms as used by us for IdeasWire. Initially the image processing algorithm were developed so that we can enhance images we captured, but now the algorithms are capable of detecting the human by its face one such algorithm is being used by the social media giant facebook in there tagging systems. It automatically provides suggestion of who that person is as soon as the photo is uploaded on there server,its not that they have a specialised team for this recg ,but there algorithm is so eficient and good.
But that’s not all! in today’s era image processing is used to detect tumor finger prints and lots of other things.
Every thing comes with a cost so does this. There are numerous of disadvantages of image processing. The setup and maintenance of image processing systems is too high ,and almost all of the image processing algorithm have a trade of that is : “Either you detect every image correctly but this will detect the false image too ,or you can develop an algorithm which does not detect false images but gives more then one answer on correct image”. The choice is yours.