Minutiae detection solutions needed for small entities' attacks

A recent video showing a flurry of sucker demons invading my body underscores the need for in-home demonic activity detection equipment that is configured to find small entities the size of those little, flying white strands of hell-fury. At least for me it does; no one else seems to be as concerned.

Floating in from the top-left...

...is a sucker demon...

...captured with unusual clarity

But, I don't think that's a lack of diligence, ignorance, complicity, culpability, surrender, stupidity, laziness, irresponsibility, stubbornness, cowardice or fear—or any of the other things I could think of that it looks like—on anyone's part. I think people just can't see them.

This video clip supports that assessment perfectly, in that I think it shows that sucker demons prefer to attack from behind and, of course, benefit from being really small and quiet and slow:



Experimenting with a multi-frequency band-pass filter of my own design, which divides an image into two user-specified frequencies, enhances (either sharpens and/or contrasts) and then combines them with the original again. This is a first step in the development of surface analysis filters similar to that already used by medical imaging devices, that step being isolating the actual surface prior to enhance its details:



The purpose is to put in the hands of non-medical professionals (i.e., the rest of us) a tool for finding cloaked and small things on surfaces that should not be there and/or are harmful. A perfect illustration of such things would be, of course, shown in the first video, and by these still frames taken from the second one:

The first step in finding small entities on surfaces...

...is by isolating the frequencies in the image showing them...

...and then enhancing those frequencies (not shown here)

As you can surmise from the filter testing video, surfaces are identified by first selecting the frequencies of the image containing the surfaces. That requires making adjustments on the part of the user, which requires ease-of-use considerations on the part of the designer (hence the testing). After that, various enhancements can be applied to sharpen and further isolate them, a step not shown in the video or still frames above.

That step will be hard to narrow down, not to mention refine. To give you an idea, here are a few of those options:

• Edge/Line detection and edge linking. Kirsch, pyramid, Sobel, Prewitt, Roberts, Laplacian, Frei-Chen, Hough transform.
• Image segmentation. Fuzzyc mean, histogram thresholding, median-cut, principal components transform/median cut, spherical coordinate transform/center split, gray level quantization, split and merge.
• Morphological filters. Binary iterative morphology, gray-scale and color erosion, dilation, opening, and closing.
• Two-dimensional fast transforms. Fourier (FFT), cosine (DCT), Haar, Walsh, Hadamard, wavelet transforms.
• Frequency domain filters. Highpass, lowpass, bandpass, bandreject, high frequency emphasis, and notch.
• Feature extraction. Binary, RST-invariant, histogram, spectral and texture object features.