Activity prediction using portable multispectral laser speckle imager

A system comprising:

• a portable multispectral laser speckle imager operatively coupled to a computing device, wherein the portable multispectral laser speckle imager further comprising a microcontroller, a transistor and a plurality of light sources, wherein the computing device comprising at least one memory storing programmed instructions;
• one or more Input/Output (I/O) interfaces;
• and one or more hardware processors operatively coupled to the at least one memory, wherein the one or more hardware processors are configured by the programmed instructions to:
• receive a plurality of multispectral image frames and a Red Green Blue (RGB) image from an imaging device, wherein each the plurality of multispectral image frames comprises a plurality of samples for activity prediction;
• extract a plurality of color component image frames, corresponding to each of a plurality of spectrums, from the plurality of multispectral image frames;
• segment the RGB image into a foreground information and a background information using a segmentation technique;
• generate a masked image based on the segmented RGB image by filling the foreground information with ones and the background information with zeros;
• generate a clustered mask image based on the masked image using a clustering technique, wherein each cluster is assigned with a corresponding value;
• create a sample region mask for each of the plurality of samples based on the corresponding cluster values associated with the clustered mask image;
• simultaneously generate a color coded activity map associated with each of the plurality of spectrums, based on the plurality of corresponding color component image frames using a color coded activity map generation technique, when creating the sample region mask;
• compute a cluster contrast value for each of the plurality of samples corresponding to each of the plurality of spectrums by multiplying the sample region mask associated with each of the plurality of samples with the color coded activity map corresponding to each of the plurality of spectrums;
• select a plurality of optimal color coded regions of interest from the plurality of color coded activity maps based on the plurality of cluster contrast values corresponding to each spectrum, wherein a color coded region of interest with a cluster contrast value greater than among the plurality of cluster contrast values associated with each spectrum is selected as one of the plurality of optimal color coded regions of interest;
• and
• generate an activity map by combining the plurality of optimal color coded regions of interest, wherein the activity map predicts the activity associated with the plurality of samples.

One or more non-transitory machine-readable information storage mediums comprising one or more instructions which when executed by one or more hardware processors cause:

• receiving a plurality of multispectral image frames and a Red Green Blue (RGB) image from an imaging device, wherein each the plurality of multispectral image frames comprises a plurality of samples for activity prediction;
• extracting a plurality of color component image frames, corresponding to each of a plurality of spectrums, from the plurality of multispectral image frames;
• segmenting the RGB image into a foreground information and a background information using a segmentation technique;
• generating a masked image based on the segmented RGB image by filling the foreground information with ones and the background information with zeros;
• generating a clustered mask image based on the masked image using a clustering technique, wherein each cluster is assigned with a corresponding value;
• creating a sample region mask for each of the plurality of samples based on the corresponding cluster values associated with the clustered mask image;
• simultaneously generating a color coded activity map associated with each of the plurality of spectrums, based on the plurality of corresponding color component image frames using a color coded activity map generation technique, when creating the sample region mask;
• computing a cluster contrast value for each of the plurality of samples corresponding to each of the plurality of spectrums by multiplying the sample region mask associated with each of the plurality of samples with the color coded activity map corresponding to each of the plurality of spectrums;
• selecting a plurality of optimal color coded regions of interest from the plurality of color coded activity maps based on the plurality of cluster contrast values corresponding to each spectrum, wherein a color coded region of interest with a cluster contrast value greater than among the plurality of cluster contrast values associated with each spectrum is selected as one of the plurality of optimal color coded regions of interest;
• and
• generating an activity map by combining the plurality of optimal color coded regions of interest, wherein the activity map predicts the activity associated with the plurality of samples.