Qfora concierge App collects customers’ PI and approves beauty and fashion styles based on this.
1. Personal color diagnosis
Qfora’s personal color type can be collected from online and offline diagnosis.
Through the camera equipped with Qfora’s concierge module, face part is separated from the image of the user.
Face part separation is a technology that performs pre-processing and background separation of video obtained from mobile and extracts facial landmark-related characteristics. Face Landmark is typically defined as face area, pupil area, nose area, lip area, eyebrow area, hair area, etc. This is within the range of visual location extraction. The technology that can extract face landmarks such as face area, pupil area, nose area, lip area, eyebrows area, hair area, and related image characteristics is Color histogram, Edge detection.
Face Landmark key points are typically first-divided using color vibration analysis and can be divided into areas of hair, face, eyes and lips.
If the value of color histogram is similar, the area can be defined graphically and accurately, depending on the Face Landmark’s relative position. Also, to extract Face Landmark accurately, you can perform a video with Template matching, Bayesian modeling, etc. to extract hair, face, eyes, lips areas.
Separate the cheeks, hair, and pupil elements of the face in the above manner and find L*a*b * of the elements.
Color synthesis techniques convert existing R, G and B images into an L*a*b * domain and apply the newly designed formula. In order to minimize the effect of the light (brightness), the value of the L was set relatively low.
𝛥𝐿,𝛥𝑎,𝛥𝑏 𝑐𝑎𝑛 𝑏𝑒 𝑜𝑏𝑡𝑎𝑖𝑛𝑒𝑑 𝑓𝑟𝑜𝑚 𝛥𝐿={1𝑚𝑛Σ.𝑚𝑖=1Σ𝐿(𝑖,𝑗)−𝐿𝑑𝑛𝑗=1} ,
𝛥𝑎={1𝑚𝑛Σ.𝑚𝑖=1Σ𝑎(𝑖,𝑗)−𝑎𝑑𝑛𝑗=1}, 𝛥𝑏={1𝑚𝑛Σ.𝑚𝑖=1Σ𝑏(𝑖,𝑗)−𝑏𝑑𝑛𝑗=1}.
With this, the total value of L,a,b can be obtained from
𝐿𝑜𝑢𝑡𝑝𝑢𝑡=𝐿(𝑖,𝑗)+0.5 X 𝛥𝐿 , 𝑎𝑜𝑢𝑡𝑝𝑢𝑡=𝑎(𝑖,𝑗)+0.5 X 𝛥𝑎 , 𝑏𝑜𝑢𝑡𝑝𝑢𝑡=𝑏(𝑖,𝑗)+0.5 X 𝛥𝑏.
Convert images of L*a*b * domains obtained through formulas into RGB domains again.
The corresponding element lab values are converted to RGB and weighted for the elements of the warm and cool compared to the standard index values of the colors. In a variety of settings, distortion of color that occurs during shooting is corrected by white balance.
IF you want more information, visit our website www.qfora.io
Thanks.
