A visual positioning apparatus, method, and non-transitory computer readable storage medium thereof are provided. The visual positioning apparatus derives an image by sensing a visual code marker in a space and performs the following operations: (a) identifying an identified marker image included in the image, (b) searching out the corner positions of the identified marker image, (c) deciding a marker structure of the identified marker image according to the corner positions, wherein the marker structure includes vertices, (d) selecting a portion of the vertices as first feature points, (e) searching out a second feature point for each first feature point, (f) updating the vertices of the marker structure according to the second feature points, (g) selecting a portion of the updated vertices as the third feature points, and (h) calculating the position of the visual positioning apparatus according to the third feature points.