computational-imaging/threedeconv.jl

A convex 3D deconvolution algorithm for low photon count fluorescence imaging

threedeconv.jl's Issues

Deconvolute Confocal Images

First, Thanks for this awesome package! I am really enjoy it using for widefield images. But i have a little question. Is this package suitable to deconvolute confocal images in Z-stack mode? If the answer is "yes", i should to have psf image generated with depth same as my image depth, and deconvolute each image in "for" cycle?

CuArray error upon calling deconvolve

Thanks for the package! Excited to try it out, but calling deconvolve causes the following error (using the example code).
Julia 1.5.3 and up-to-date packages with CUDA.

MethodError: no method matching zero(::Type{CuArray{Float32,N} where N})
Closest candidates are:
  zero(!Matched::Type{Pkg.Resolve.FieldValue}) at /home/.../Src/julia/julia-1.5.3/usr/share/julia/stdlib/v1.5/Pkg/src/Resolve/fieldvalues.jl:38
  zero(!Matched::Type{Dates.Time}) at /home/.../Src/julia/julia-1.5.3/usr/share/julia/stdlib/v1.5/Dates/src/types.jl:406
  zero(!Matched::Type{Dates.DateTime}) at /home/.../Src/julia/julia-1.5.3/usr/share/julia/stdlib/v1.5/Dates/src/types.jl:404
  ...

Stacktrace:
 [1] zeros(::Type{CuArray{Float32,N} where N}, ::Tuple{Int64,Int64,Int64}) at ./array.jl:526
 [2] ThreeDeconv.ADMMstate(::Tuple{Int64,Int64,Int64}, ::Array{Tuple,1}) at /home/.../.julia/packages/ThreeDeconv/vtLlk/src/deconvolution/admm.jl:78
 [3] initialize_optimizer(::ThreeDeconv.ADMM, ::ThreeDeconv.StackedLinearOperator, ::ThreeDeconv.var"#58#65"{CuArray{Complex{Float32},3},ThreeDeconv.LinearOperator,ThreeDeconv.LinearOperator,CuArray{Complex{Float32},3},CuArray{Complex{Float32},3},CuArray{Complex{Float32},3},CuArray{Complex{Float32},3},CuArray{Complex{Float32},3},CuArray{Complex{Float32},3}}, ::ThreeDeconv.var"#60#67"{Float32,Float32,Float32}) at /home/.../.julia/packages/ThreeDeconv/vtLlk/src/deconvolution/admm.jl:103
 [4] setup_optimizer(::ThreeDeconv.ADMM, ::Array{Float32,3}, ::Array{Float32,3}, ::Float32, ::Float32, ::Float32) at /home/.../.julia/packages/ThreeDeconv/vtLlk/src/deconvolution/setup_admm.jl:128
 [5] deconvolve(::Array{Float32,3}, ::Array{Float32,3}, ::Float64, ::Float64, ::Float64, ::ThreeDeconv.ADMM; options::ThreeDeconv.DeconvolutionOptions) at /home/.../.julia/packages/ThreeDeconv/vtLlk/src/deconvolution/deconvolve.jl:88
 [6] top-level scope at In[25]:4
 [7] include_string(::Function, ::Module, ::String, ::String) at ./loading.jl:1091

[WIP] Support non-square images

cuFFT always fail to allocate GPU or CPU memory

When I run ThreeDeconv.deconvolve in the first time, it work well.
But at second time, error occur:
CUFFTError(code 2, cuFFT failed to allocate GPU or CPU memory)
Then I run same command against, no error!!! I am confused why GC.gc() don't release GPU memory.

I run Julia 1.2 in Fedora 30 with a NVIDIA 2080Ti.

What should I do?

ERROR: could not find project file in package at https://github.com/computational-imaging/ThreeDeconv.jl.git

When I hit the key ] in the Julia REPL to ender the Pkg REPL-mode, and run
pkg> add https://github.com/computational-imaging/ThreeDeconv.jl.git
There is an error:
ERROR: could not find project file in package at https://github.com/computational-imaging/ThreeDeconv.jl.git
I am not familiar with Julia language, does this affect use？

Register Registrator?

@JuliaRegistrator register()

Please update so this works on current Julia 1.7

Please update so this works on current Julia 1.6. Failure:

Hit the key ] in the Julia REPL to ender the Pkg REPL-mode, and run
pkg> add https://github.com/computational-imaging/ThreeDeconv.jl.git

resulted in error ... for package [CUDA]
Screen shot attached.