From Python directly, you could use this simple IGV wrapper from @brentp. Copied from the example usage of that script:

#example usage:

    >>> igv = IGV()
    >>> igv.genome('hg19')
    'OK'

    >>> igv.load('http://www.broadinstitute.org/igvdata/1KG/pilot2Bams/NA12878.SLX.bam')
    'OK'
    >>> igv.go('chr1:45,600-45,800')
    'OK'

#save as svg, png, or jpg
    >>> igv.save('/tmp/r/region.svg')
    'OK'
    >>> igv.save('/tmp/r/region.png')
    'OK'

# go to a gene name.
    >>> igv.go('muc5b')
    'OK'
    >>> igv.sort()
    'OK'
    >>> igv.save('muc5b.png')
    'OK'

# get a list of commands that will work as an IGV batch script.
    >>> print "\n".join(igv.commands)
    snapshotDirectory /tmp/igv
    genome hg19
    goto chr1:45,600-45,800
    snapshotDirectory /tmp/r
    snapshot region.svg
    snapshot region.png
    goto muc5b
    sort base
    snapshot muc5b.png

# Note, there will be some delay as the browser has to load the annotations
# at each step.

Download the IGV : You can automate it without manually opening the IGV: If starting file is bed file, use "bedToIgv" from bedtools to generate the igv_batch.file for snapshots:

For example in the Linux: using indexed bam file

bash igv.sh input.bam -g hg19 -b igv_batch_script.txt

#sample igv_batch_script.txt ##

new
genome hg19
load input.bam
snapshotDirectory /path/to/igv_screenshots
goto chr19:59418052-59418053
sort base
collapse
snapshot chr19_59418052-59418053.png
goto chr1:15680529-15680532 chr3:5680521-5680533 
sort base
collapse
snapshot chr1_15680529-15680532.chr3_5680521-5680533.png
exit

If you install and setup GBrowse2 you can programmatically generate images for each region (see here). I have used GBrowse2 before to do this with next generation sequencing, it's quite nice because the height of the image is not limited and you can add additional annotations such as coverage plots and gene annotations.

most genome browsers have scripting capability [IGB] and [IGV].
GBrowse2 also allows linking and by this scripting.

If you want something really small:

1. download the two files from lookseq [lookseq]
2. and compile them: g++ renderimage.cpp -O3 -o renderimage -lpng -L . -lbam
3. create image with: ./render_image --bam=mybam.bam --options=snps,pairs,arrows,single,faceaway,inversions,linkpairs,colordepth --ref=myref.fa --region="2L:1-200000" --png=2L.a.png

This way you can quickly create small images with the reads from your bam file with many viewing options. like this [example] (click on one button on the right).

I don't really have a definitive answer, I asked in similar terms a while ago but I think it would be great to have a definitive answer to the question:

Quick Visual Inspection Of Mapping Images For A List Of Regions

cat peaks.txt | while read i; do echo -e "g$i\n." | samtools tview my.bam my.ref.fa.gz; done

The "." at the end is to switch of the dots in the tview configuration. I use Ctrl+C to go to the next region.

Is there a better way to do this? Maybe something that generates image files?

@2184687-1231-83: You can use tview for automatized analysis, using the samtools Version: 1.3.1 For instance, with this Bash script

while read coor; 
do 
samtools tview -d T -p $coor <<Indexed Bam file>>  <<genome>> > $coor.txt 
done < Test_coordinates.txt

The Test_coordinates.txt file could be chr9:86584727 chr15:89155087 chr19:4770712