Aiming at the problem of music reading efficiency for visually impaired people, this paper proposes a demand-oriented algorithmic framework for Braille music score layout optimization. In the image recognition layer, the Histogram of Orientation Gradients (HOG) algorithm is adopted to realize Braille dot matrix feature extraction, and the geometric invariant feature vectors are generated through the steps of grayscaling, Gamma correction and gradient calculation. In the system conversion layer, the formal conversion model from MusicXML to Braille symbols is constructed, and the precise mathematical mapping relationship from music elements to Braille ASCII code is established by defining the composition function, rule function and conversion function, and the semantics of the music score is parsed based on the DOM tree structure. At the interaction design level, the operational efficiency of QWERTY and alphabet layout is compared. In order to verify the system performance, seven pieces of music with different complexity levels were selected for testing. Among them, 97.07%-99.12% were in Level 1 Braille and 96.25%-98.08% were in Level 2 Braille. The highest percentage of note conversion errors (88.2% for Level 1 and 89.9% for Level 2) was due to the failure of complex chord and ornamentation processing. The conversion time was linearly correlated with the scale of the score, with an average processing speed of 6.5 bars/ms for Level 1 Braille, Score D: 308ms/484 bars, and a 90% increase in time for Level 2 Braille, Score B: 469ms/152 bars. Braille sensing unit signal test shows that the signal peak variation under different users’ touch is <15%, and the peak pattern consistency reaches 92%, confirming the interaction universality. The study provides a full-chain solution for Braille sheet music reading terminals, which significantly improves the efficiency of music reading for the visually impaired.
