QMap - Image Similarity detection

Image similarity detection is a useful tool we created. It compares the histograms of two images and detects the position of the color range on the image. At the same time, it also detects the similarity rate of color and structure, and finds the difference area in the geometric position of the two images. Since the given "color similarity ratio" is a comparable parameter, it provides an effective method for quickly screening images.

Now according to the software function interface, the introduction is as follows:

1. Click the right mouse button to select the image

    Two images are arbitrarily selected and displayed in the dialog box, and the system automatically counts the histogram. Use the mouse or keyboard direction keys, move the slider of the sliding bar, select the color value in the histogram, and use it with the Shift key to select the color interval.

2. Under each channel selection box, the current color value or interval, the total number of points in the interval, and the percentage of these points to all pixels are displayed respectively. 

3. How to display results

a) With background + information simplification: On the original image, the extracted information is replaced with a color for quick verification. b) No background + single information: The extracted information is replaced with one color, and the other information is replaced with the background color to obtain a binary image, which is convenient for data filing and other applications. c) No background + information extraction: The extracted information pixels are displayed in the primary color, and the non-information pixels are displayed as the background color, and the effect is very clear. Converting the above three methods can improve the analysis efficiency. 

4. Information color and background color of the result graph are optional

5. Histogram check

    Click the "Histogram check" button, when "Image 1" is selected at the top of the dialog box, a new window will pop up to display the processing result image of "Image 1". Similarly, when "Image 2" is selected, the processing result image of "Image 2" pops up, and the two images are aligned and viewed at a glance.

In addition, in order to achieve the purpose of quick viewing, you can select the "Reported" option to decide whether to pop up a statistical report every time.

6. Detect similarity between two images

    a) Use different algorithms to detect the similarity between different types of images, so as to achieve the purpose of fast processing and judgment by the computer. There are currently 3 algorithms available.

b) The size of the "Test area“ and ”Sampling area" can affect the accuracy of the analysis. A larger training area and a smaller sampling area can improve accuracy, but increasing operations will affect speed. c) "Show dissimilarity" option provides two images that are different after operation. Selecting "Show Dissimilarity" means that the resulting images show different parts of both images.

7. Click the "Detect similarity" button

    When "Image 1" is selected, the color similarity rate, maximum difference and structural similarity rate of the two images will be reported, and then the processed "Image 1" structural difference image will pop up. Similarly, the structural difference image of "image 2" is obtained. Two difference images can be displayed on the same screen, which is clear at a glance.

Back to www.ccqmap.com