Method of utilizing wide-angle image capturing element and long-focus image capturing element for achieving clear and precise optical zooming mechanism

1. A method of utilizing a wide-angle image capturing element and a long-focus image capturing element to achieve clear and precise optical zooming when controlling an electronic device to simultaneously and respectively capture a wide-angle image and a long-focus image of a same spot comprising the steps of:

reading the wide-angle image and the long-focus image, and performing exposure and white balance adjustment processes to the wide-angle image and the long-focus image respectively, so as to enable the exposure and the white balance of the wide-angle image and the long-focus image to be in consistency;

performing an image matching algorithm to the wide-angle image and the long-focus image respectively, so as to enable each pixel on the long-focus image to match with each corresponding pixel on the wide-angle image; wherein the image matching algorithm calculates and obtains an image ratio between the long-focus image and the wide-angle image according to hardware parameters of the long-focus image capturing element and the wide-angle image capturing element, retrieves a region from the wide-angle image that corresponds to the long-focus image according to the image ratio, and then calculates and obtain dense optical flow field of an offset of each pixel on the long-focus image with respect to each corresponding pixel on the wide-angle image by using an optical flow estimation; and

performing an image morphing fusion process to the corresponding pixels on the long-focus image and the wide-angle image, so as to generate a new wide-angle image; wherein the image morphing fusion process performs an offset deformation to the pixels on the long-focus image T(x+u, y+v) according to the dense optical flow fields of the offsets (u, v) of the pixels on the long-focus images T(x+u, y+v), so as to enable the pixels on the long-focus image T(x+u, y+v) to be deformed to match with the corresponding pixels on the wide-angle image W(x, y), and then perform a fusion process to the long-focus image T(x+u, y+v) and the wide-angle image W(x, y), in accordance with the following formula, so as to generate the new wide-angle image W NEW :

W NEW =(1−λ) T ( x+u,y+v )+λ* W ( x,y )

wherein λ is a weight factor and

wherein, when pixels on a deformed long-focus image T(x+u, y+v) are clearer than the corresponding pixels on the wide-angle image W(x, y), taking λ=0, otherwise, taking 0<λ<1.0.

2. The method according to the claim 1 , wherein the optical flow estimation is performed in accordance with the following formula based on a Horn-Schunck optical flow estimation model, so as to obtain the dense optical flow field through converting an optical flow estimation into a process for obtaining the smallest error between the corresponding pixels on the deformed long-focus image T(x+u, y+v) and the wide-angle image W(x, y):

min E ( u,v )=∫ Ω [(φ( W ( x,y )− T ( x+u,y+v ))+α*ψ(|∇ u|,|∇v |)] dxdy

wherein, φ(W(x, y)−T(x+u, y+v)) is a data item representing the gray value conservation constraint for enabling each corresponding pixels on the wide-angle image W(x, y) and the long-focus images T(x+u, y+v) to be the same as far as possible, φ is a first error function; α*ψ(|∇u|,|∇v|): is a smooth item representing the optical flow smoothness constraint for enabling the offset between every two adjacent pixels on the wide-angle image W(x, y) and the long-focus image T(x+u , y+v) not to be much different, α is a weight factor for controlling the weight between the data item and the smooth item, ψ is a second error function, and ∇u, ∇v are modulus lengths of the offset gradients.

3. The method according to the claim 2 , wherein at least one of the first and second error functions φ and ψ is a quadratic function.

4. The method according to the claim 3 , wherein the presumption of the optical. flow estimation is that brightness differences between the corresponding pixels on the long-focus image T(x+u, y+v) and the wide-angle image W(x, y) are constant.

5. The method according to the claim 4 , wherein the presumption of the optical flow estimation is that motion differences between the corresponding pixels on the long-focus image T(x+u, y+v) and the wide-angle image W(x, y) are small.

6. The method according to the claim 5 , wherein the presumption of the optical flow estimation is that motion vectors of the corresponding pixels on the long-focus image T(x+u, y+v) and the wide-angle image W(x, y) are the same.